Types of Fillet Weld: The Complete Guide to Profiles, Joints, Sizing, Codes, and Technique

Table of Contents

Why Fillet Welds Run the Metalworking World (And Why ‘Types’ Matter)

Marcus was three weeks into his first fabrication job and already proud of himself. He had spent the afternoon running T-joint after T-joint, laying beads he thought looked like frosting on a cake. Then Dale, a 30-year inspector with a flashlight and no patience, walked the bench. “That one’s convex,” he said, tapping the first. “That one’s flat. That one’s concave, and it’s a sixteenth small. Two of these don’t match the drawing size.” Marcus had made one weld in one position, and Dale had just found three profiles and an undersized joint. That single bench walk is the whole reason this guide exists.

Here’s the thing: “types of fillet weld” isn’t one neat list. It’s a family of classifications that stack on top of each other.

Quick Note: A fillet weld is a weld of approximately triangular cross-section joining two surfaces that meet at approximately right angles to each other — usually at a tee, lap, or corner joint. That makes it fundamentally different from a groove weld, which fills a gap or prepared bevel between square or beveled edges. AWS A3.0 is the standard that codifies this definition.

Those classifications fall into five practical buckets:

Joint configuration — tee, lap, corner, and edge joints
Face profile — convex, concave, and flat (including the 45-degree mitre fillet)
Continuity — continuous, chain intermittent, staggered intermittent, stitch, and tack
Number of sides — single-sided versus double-sided
Welding process — SMAW, GMAW, FCAW, GTAW, SAW, and robotic automation

If you typed “types of fillet weld” into a search bar, you probably want four answers: what the categories are, how to recognize each one on a drawing or in the field, which type belongs on which job, and what the codes actually allow. That’s the roadmap here. We’ll cover anatomy first, then profiles and joint configurations, then continuous versus intermittent and single versus double-sided welds, then sizing and symbols, strength and fatigue, defects, and finally how to choose and run the right fillet.

Why does this matter? Because the fillet weld is the workhorse of the trade. Industry estimates commonly put fillet welds at roughly 70–80% of all welds in structural steel fabrication, making them the most common weld type by volume — the gusset on a column, the bracket on a beam, the rail on a frame. Get the profile, size, and continuity right and those connections are strong, economical, and code-compliant. Get them wrong and you’re grinding, re-welding, and explaining yourself to an inspector like Dale.

Consider the money. The same 1/4 in. fillet can cost roughly three times more as a double-sided weld than a single-sided one — double the arc time, double the consumables, double the inspection. And that same 1/4 in. fillet behaves very differently in fatigue depending on whether it’s continuous or intermittent, convex or concave, single or double-sided. Identical on the drawing, completely different in service.

Fillet weld types also dictate how you actually run the bead. A convex fillet in the flat position is a different animal from a slightly weaved concave fillet, and both differ again from a staggered intermittent pattern on a long rail. If you want to build fundamental consistency before worrying about classifications, our guide to welding practice is a solid starting point, and learning to read bead appearance will help you catch profile problems early — see good and bad MIG welds.

By the end of this guide you’ll be able to look at a fillet, name its type, check it against the symbol, and know whether it belongs there. Let’s start with the anatomy that every profile, size, and strength calculation grows out of.

Fillet Weld Anatomy: Legs, Throat, Face, Root, Convexity, and Concavity Explained

Before you can judge whether a fillet weld is the right type, you need the vocabulary to describe it. A fillet weld is a triangle of deposited metal tucked into the corner where two surfaces meet — and every term on a welding drawing or inspection report maps to a specific feature of that triangle.

The Root, Toes, and Face

Start in the corner. The root is where the two base-metal surfaces meet — the deepest point of the joint. From there the weld metal rises up against each plate.

The toes are the lines where the weld face blends into the base metal; a typical fillet has two. This is where undercut, overlap, and lack of fusion begin, so toes get close attention during inspection.

The face is the exposed surface running from toe to toe. Its shape — called the profile — is where convexity and concavity live.

Legs: The Measured Dimensions

The legs are the two fused surfaces of the weld — the distance the weld extends along each plate. On an equal-leg fillet in a 90-degree T-joint, both legs are equal, and this is the number specified on drawings and in AWS D1.1 callouts. Leg length is what you measure with a fillet gauge; it is not what you calculate strength from.

Throat: Theoretical, Actual, and Effective

Strength runs through the throat. Picture the inscribed right triangle: the hypotenuse runs along the face, and the two legs run along the plates.

Theoretical throat — the perpendicular distance from the root to the face of the inscribed triangle. For an equal-leg, 90-degree fillet, theoretical throat = 0.707 × leg size.
Actual throat — the real measured throat of the finished weld, including any convexity.
Effective throat — the value the designer credits for strength. On flat and convex welds it equals the theoretical throat; on a concave weld it equals theoretical throat minus the depth of concavity.

The inscribed triangle is worth picturing literally. Lay a right triangle inside the weld cross-section so its legs sit flush with the two plates and its hypotenuse connects the toes — that triangle is what the design assumes is filled with sound metal. Anything beyond it, like convexity, is extra credit. Anything missing inside it, like concavity or lack of fusion, is a debit on capacity.

Run the math. A 1/4 in leg fillet has a theoretical throat of approximately 0.177 in — that is 0.707 × 0.25. A 3/8 in leg produces roughly 0.265 in.

Short Note: never assume leg size equals throat size. Drawings and AWS D1.1 callouts specify leg dimension for equal-leg fillets, but strength calculations use the effective throat.

Because effective throat drives capacity, two welds with identical 1/4 in legs can carry different loads. Grind a concave profile too deeply and you subtract from the only dimension the code credits.

Convexity, Concavity, and the Field Reality

A convex fillet bulges outward. That extra reinforcement forgives slightly gapped root fit-up, but it raises stress concentration at the toes — which is why AWS D1.1 visual criteria cap maximum convexity and reject undersize welds.

A flat (mitre) fillet is essentially the inscribed triangle. A concave fillet dips inward, smoothing the load transition into the base metal and lowering the fatigue notch, but every bit of concavity is subtracted from the effective throat. If profile control is giving you trouble, the bead-reading habits in welding practice and our breakdown of good and bad MIG welds will help you spot these features by eye.

Picture a fabricator who runs a grinder along a finished fillet to make it look tidy. On a convex weld, grinding removes material; the leg may still measure acceptably, yet the throat — and the effective throat — has quietly shrunk below what the drawing required.

Master these terms and every other fillet weld decision — profile, sizing, and inspection — becomes measurable instead of guesswork.

Fillet Weld Profiles: Convex, Concave, Flat, and the 45-Degree Mitre Fillet

When most people say “types of fillet weld,” they mean profile — the shape of the weld face running between the two toes. Profile is set by process, technique, and travel speed, and it changes how a joint behaves under load far more than most fabricators realize. Four profiles cover nearly everything you will meet: convex, concave, flat (mitre), and the asymmetrical fillet. The AWS D1.1 structural welding code treats profile as a measurable acceptance criterion, not a cosmetic preference, which is why an inspector can reject a weld that “looks strong.”

Convex Fillet: The Default From Stick and Flux-Cored

A convex fillet bulges outward, a rounded bead piled highest at the centerline. Gravity and surface tension shape it, which is why flux-cored and stick welding produce it almost automatically at comfortable settings and travel speeds. It is easy to deposit, forgiving of poor root fit-up, and fills gaps that would sink a flatter bead. The trade-off is a sharp geometric notch where the bead meets the base metal. That toe is a stress raiser, and when convexity exceeds code limits it counts as excessive reinforcement — a defect in several acceptance criteria precisely because it lifts the fatigue notch.

Concave Fillet: Dished Face, Lower Stress Concentration

Push the puddle’s edges until the face dishes inward and you have a concave fillet. Welders get there with a weave, multiple stringers, or a controlled grinding pass. The smooth toe transition lowers stress concentration, so fatigue-loaded structures — bridge details, crane girders, cyclically stressed machinery — favor it. The trap is grinding: remove metal below the specified throat and you have created an undersized weld with perfectly pretty geometry. Concave is a profile you must measure, not eyeball. Building the hand skill to control this consistently is pure deliberate welding practice, not luck.

Flat (Mitre) Fillet: The Profile Your Calculations Assume

With equal legs and a face that runs as a straight line from toe to toe, the profile is flat — often called a mitre fillet because the face sits at roughly 45 degrees. This is the ideal assumed by the standard throat formula, t = 0.707 × leg. Real welds only approximate it, but a flat face is what pressure-vessel and machinery baseplates, gusset connections, and machine bases are usually drawn around, because the face lands flush without extra reinforcement.

Asymmetrical Fillets and Skewed Tees

Equal legs are an assumption, not a rule. When one member is thinner, or the joint angle is not 90 degrees, you get an asymmetrical fillet — unequal legs sized so the throat lands where the load needs it. Skewed T-joints are the classic case: as the angle opens or closes, the effective throat no longer equals 0.707 × leg, and the obtuse side loses throat fast. AWS D1.1 carries special provisions for skewed T-joints, allowing an increased leg size so the required throat is actually achieved — an allowance welders who “weld it like a normal tee” routinely miss.

Quick Note: An engineer specifies a concave fillet on a crane runway bracket. The fabricator ships a full convex bead — plenty of metal, arguably stronger-looking. The inspector flags it anyway, because on a cyclically loaded bracket the sharp toe notch becomes a fatigue crack initiation site. Profile is a design instruction, not a style choice, and no grinder session fixes it later without risking the specified throat.

The takeaway: profile is engineered geometry. Match the specified face — convex, concave, flat, or asymmetrical — and treat every toe as a potential crack origin.

Fillet Welds by Joint Configuration: T-Joints, Lap Joints, Corner Joints, and Edge Joints

Profile tells you what a fillet weld looks like. Joint configuration tells you how the plates meet — and whether a fillet is allowed at all. Nearly every fillet you run falls into four families: T-joints, lap joints, corner joints, and edge joints.

The T-Joint: The Classic Fillet Application

Two plates meet at roughly 90 degrees, one butting into the face of the other to form a T, and the fillet sits in that inside corner. Load can travel through one side or both, so you can run a single-sided tee or lay fillets along both toes.

This is the default geometry for brackets, stiffeners, gussets, beam webs, and base plates. The corner gives the puddle an accessible, self-supporting trough, and the tee transfers load across the throat in shear rather than demanding penetration through the plate — exactly the duty a fillet is built for. AWS D1.1 treats the tee as the benchmark fillet joint.

Lap Joints: Adjustability and Sealing

One plate overlaps another, and the fillet runs along the exposed edge of the top plate. Weld one edge or both — the double-sided version dramatically increases the effective throat because the load path crosses two fillets instead of one. Lap joints forgive fit-up, allow adjustment before tacking, and seal well, which is why they dominate sheet metal and tank work. Weld both edges when the joint carries real load or must be leak-tight.

Corner Joints: Tee vs Corner in Plain Terms

A tee has one plate’s edge landing on another plate’s flat face. A corner joint has both plates meeting at their edges to form an L or a box. On a tee the fillet lives inside; on a corner it can sit inside or outside. A closed corner — plates butted tightly at 90 degrees — accepts an outside fillet only when fit-up is tight, because gaps let the arc blow through and starve the root.

Edge Joints: Why a Fillet Often Fails

Two plates laid edge-to-edge in the same plane. Abutted with square edges, there is no corner for the puddle to sit in, so a fillet cannot be applied — the arc runs across two surfaces with no common root. To fillet an edge joint you need flanged edges turned up into two small fillets, or enough thickness plus a beveled prep. Add the bevel and you have really specified a groove weld, and TWI’s technical FAQs agree: edge joints belong to groove welds unless the geometry changes.

Quick Note: fillet welds apply to T-joints, lap joints, corner joints, and — with a flange or enough thickness — edge joints. If two plates are merely abutted edge-to-edge with no flange or bevel, no fillet will join them.

Practical selection guidance:

T-joint fillet: structural brackets, stiffeners, gussets, web-to-flange connections.
Lap joint: work needing adjustability and sealing — tanks, sheet-metal assemblies, repair plates.
Corner fillet: frames, boxes, and enclosures with tightly fitted corners.
Edge joint: avoid fillets unless you flange the edges or add backing; otherwise specify a groove weld.

A trailer-frame job shows why. Two plates are butted edge-to-edge, the welder turns up the heat, and a fillet is dragged down the seam. Nothing fuses — the puddle rolls over two separate surfaces, and the joint pops apart under load. Redesigned as a proper corner joint, with the plates meeting at 90 degrees, the same filler suddenly has a throat to bite into.

Note: fillet welding is not structurally equivalent to a full-penetration groove weld. Fillet welds are generally partial-penetration joints and do not develop the full tensile strength of the base metal across the joint. When a connection must carry full plate capacity — a tension splice or moment connection — the drawing demands a groove weld instead.

Read the joint before you strike an arc. Deliberate welding practice on these four configurations builds more useful skill than chasing bead appearance, and our guide to MIG welding at home covers the setup that makes each repeatable.

Continuous vs Intermittent Fillet Welds: Chain, Staggered, Stitch, and Tack Patterns

Profile and joint configuration decide how a fillet weld looks. Continuity decides how much it costs and how badly the assembly warps. This is where estimators win bids — and where careless drawings quietly ruin parts.

Continuous Fillet Welds: Uninterrupted From Start to Stop

A continuous fillet runs the full length of the joint with no intentional breaks. It is the default whenever the connection must be watertight, resist corrosion, survive fatigue cycling, or transfer the full design load. Tank seams, pressure-retaining connections, and anything exposed to weather or wash-down usually demand it. A leaking tank seam is proof that continuity is a function, not a preference.

Continuity also spreads stress evenly. Any gap in a weld creates a stress riser — a spot where cracks like to start. Where fatigue governs, a continuous fillet is often the cheapest insurance you can buy.

Intermittent Fillet Welds: Weld, Gap, Repeat

An intermittent fillet weld breaks the joint into measured segments. The drawing gives a length and a pitch: the pitch is the distance from the start of one segment to the start of the next. So 3-9 means a 3 in long weld on 9 in centers — 3 in of weld, 6 in of gap.

Intermittent welds save filler metal, arc time, and heat input, which reduces distortion on thin material. They are only allowed where the joint does not need full-length transfer of load. For thin-gauge work, dialing in travel speed and voltage matters as much as the pattern — see our guide to MIG welding at home.

They are not a free lunch. AWS D1.1 requires each segment to meet a minimum length — commonly at least 1-1/2 in — and to be positioned so the effective length requirements of the joint are satisfied. Short, stubby segments cool too fast, fail to fuse, and simply do not count. AISC sets a similar floor for structural steel.

Chain vs Staggered Intermittent Fillets

On a T-joint welded from both sides, the two rows can relate to each other in two ways:

Chain intermittent fillet: the segments on opposite sides line up with each other, like links in a chain.
Staggered intermittent fillet: the segments on one side are offset by roughly half a pitch from the other side.

Chain welds are easier to lay out and inspect. Staggered welds balance heat input across the joint, which is why they show up on thin panels.

Quick Note: Staggered segments are usually offset by half a pitch, so the gaps on one side sit behind the welds on the other. That symmetry is what keeps thin sheet from pulling and oil-canning.

Stitch Welds and Tack Welds

“Stitch weld” is shop slang for a short intermittent weld, often used for fit-up and light fabrication. It is not a formal code term, so specify length and pitch on the drawing rather than writing “stitch it.”

Tack welds hold parts in alignment until the real weld is made. They are not structural unless specifically designed and inspected as such. Many codes require tacks to be re-melted into the final weld, so a cold, cracked, or slag-covered tack becomes a defect buried in the finished joint. For a clean start on any process, see our guide to using a wire welder and this welding practice breakdown.

Real-World Calls

On a trailer or rail car side panel, staggered intermittent welds are the classic distortion fix: thin sheet, lots of linear feet, and minimal load. On a beam stiffener, the joint transfers shear, so continuous fillets are mandatory. Specifying intermittent welds where corrosion or fatigue governs is the classic mistake — crevices between segments trap moisture and act as crack starters.

Single-Sided vs Double-Sided Fillet Welds: Symmetry, Load Path, and Distortion

Two welds can carry the same nominal leg size and still behave completely differently, because the side count changes the load path. That is the entire story behind single-sided versus double-sided fillets.

What “Side Count” Actually Changes

A single-sided fillet deposits weld on one face of a T-joint or lap joint only. The load now travels through one leg, and the force line sits off-center relative to the joint’s neutral axis. The result is an eccentric load: the weld is pulled in shear and simultaneously bent out of plane, and the joint tends to rotate under load. That rotation is not a defect you can grind away — it is built into the geometry.

A double-sided, or balanced, fillet puts weld on both sides. The two throats roughly center the load path, the eccentricity largely cancels, and the joint resists rotation far better.

Note: On thick, restrained joints, a double-sided fillet welded in a single pass each side can be more economical and more symmetrical than one large single-sided fillet. Two modest passes beat one oversized bead on both cost and distortion control.

When Single-Sided Is Fine — and When It Is Not

Single-sided is acceptable when access only permits one side, when the load is light and predominantly shear, or when the part is temporarily tacked and the real weld comes later. It is a poor choice when the joint sees significant moment, fatigue cycling, or when you need maximum joint efficiency. In those cases the code-required path is a balanced double-sided weld. If you are building up your judgment on this, the same instinct applies across processes — it is the reasoning covered in welding practice.

One caution worth stating plainly: doubling the number of fillets doubles the weld metal and the arc time, but it does not double capacity. Strength is governed by effective throat and load path, not by counting beads.

The Distortion Story, Mechanically

Weld shrinkage is unavoidable, and it is one-sided heating that wrecks squareness. Weld the top of a T-stiffener first and the contraction pulls that flange toward the weld, tipping the web out of square. Weld the bottom in the same direction and you add to the error instead of opposing it.

Picture a long T-stiffener with fillets run from the same end on both sides. Heat accumulates at the starting end, and each pass drags its shrinkage zone along behind it. The stiffener curls like a banana — a real shop headache that no amount of clamping fixes after the fact.

Technique Fixes That Work

Alternate sides: split the deposit left and right so shrinkage forces oppose.
Back-step: run short beads opposite the travel direction so each new start cancels the previous end’s pull.
Balanced heat input: keep amps and travel speed consistent on both sides.
Fixture or clamp: constrain the joint while it cools, not while it is hot.
Low-distortion processes: higher-deposition, lower-heat options such as robotic submerged arc welding limit per-pass heat on long runs.

AWS D1.1 and the AISC specification both treat balanced welding as a distortion-control method rather than a preference; the AISC and AWS standards pages are the places to confirm the clause for your project. Get the sequence right and the joint stays square; get it wrong and you spend the afternoon with a torch and a hydraulic ram.

Fillet Weld Sizing, Symbols, and Drawings: Leg vs Throat and Minimum/Maximum Sizes

A fillet weld on a drawing is a small right triangle sitting on a reference line. Everything you need to build that weld correctly — how big, how long, which side, and how often — is packed around that triangle. Read it wrong and you can build a joint that looks fine but is undersized for its load.

How to Read the Fillet Weld Symbol

The triangle points toward the joint member that gets the weld. Size goes to the left of the triangle. For an equal-leg fillet, that number is the leg length, not the throat. Length and pitch go to the right, so “3/16 (6) 12” means a 3/16 in leg, 6 in long, spaced 12 in on center.

Arrow side — weld goes on the side the arrow touches.
Other side — weld goes opposite the arrow; the symbol sits above the dashed reference line.
Both sides — symbols appear on both lines.
Field weld flag — weld is made on site, not in the shop.
All-around circle — weld continues around the entire joint.
Tail — holds process or specification notes, like “GMAW” or a WPS number.

Leg Size vs Throat Size

Leg size is the number a fabricator measures with a fillet gauge. Throat is the critical dimension governing strength, and for a 45-degree equal-leg fillet the theoretical throat is 0.707 times the leg, per TWI’s technical FAQs. Some drawings, especially skewed joints where the two members don’t meet at 90 degrees, specify effective throat directly because the leg-to-throat math no longer holds. If you’re unsure which figure is called out, check the drawing notes before you cut or weld. For anyone building skills on thinner work, this is the kind of detail worth drilling in welding practice.

Minimum Fillet Weld Size (AWS D1.1)

Under AWS D1.1, the minimum leg size rises with the thickness of the thinner part being joined. The logic is heat input: a small bead on thick, cold steel quenches fast and can crack.

Thinner part thickness	Minimum fillet leg
Up to 1/4 in	1/8 in
Over 1/4 to 1/2 in	3/16 in
Over 1/2 to 3/4 in	1/4 in
Over 3/4 in	5/16 in

Maximum Size Along an Edge

You can’t just crank the size up. For material under 1/4 in thick, the maximum fillet along an edge generally equals the material thickness. For 1/4 in and thicker, it’s generally the thickness minus 1/16 in — the base metal edge has to stay intact to fuse properly. That 1/16 in deduction preserves a shelf of base metal so the arc fuses into the edge instead of washing over it. Drawings can override this.

Minimum Length Rules

End-loaded fillets are the ones carrying load through their ends. They generally need an effective length of at least four times the nominal size. If shorter, only a fraction of the specified size counts toward strength — a shortcut that quietly robs capacity. The rule exists because stress concentrates at the ends; a short, fat bead simply can’t develop full strength along a stubby length.

Quick Note: A detailer once marked a 3/16 in fillet where 1/2 in plate joined. 3/16 in sits below the D1.1 minimum of 1/4 in for material over 1/2 to 3/4 in. The fix is simple — bump the leg to 1/4 in — but catching it before fabrication saves a rework.

Short Note: Verify leg size with a fillet gauge rather than eyeballing it. A 1/16 in shortfall can be the difference between passing and failing inspection.

None of this matters if your process settings fight you. Getting consistent leg and throat on production work — especially at high deposition like robotic submerged arc welding — depends on dialing in travel speed, wire feed, and tip condition, which is why clean, reliable maintenance pays off.

Strength, Stress, and Fatigue: How Fillet Welds Carry Load and What the Codes Require

When a fillet weld fails, it rarely does so because the filler metal snapped in two under a single, clean pull. More often, the failure comes down to geometry. The humble fillet is not a block of metal, but an angled connection, and that angle is what dictates how it carries stress. For this reason, engineers design fillet welds as loaded in shear on the effective throat plane, regardless of the direction of the applied load. Even if the member is in tension, the triangular cross-section converts the majority of that force into shear along its narrowest resisting plane.

This assumption is the backbone for sizing a weld. Under the American Institute of Steel Construction (AISC) provisions used in building design, the nominal shear strength of a fillet weld is typically taken as a fraction of the electrode’s tensile strength, most commonly 0.60 × F_EXX, applied to the effective throat area. What makes this more nuanced is direction. AISC applies a directional strength increase factor of 1.0 + 0.50 sin^1.5 θ, which credits fillet welds loaded transversely (perpendicular to the weld axis) over those loaded longitudinally (parallel to the weld axis). The reason is simple: when loaded transversely, the force is more favorably oriented relative to the throat, allowing the same leg size to resist more load. These calculations may seem theoretical, but they are what keep brackets, frames and beams standing under static loads.

The Reality of Cyclic Loading

Static strength can be more than enough, yet a weld can still crack after years of use. This is where fatigue takes over. Fillet welds are notorious fatigue weak points due to their toe geometry, which creates a sharp stress concentration where the weld meets the base metal. As a result, fillet welds generally fall into the lower AISC fatigue stress categories, meaning cyclic loading can govern the design even when a quick static check passes. This is why the same joint can hold a dead weight indefinitely, but fail under repeated vibration from equipment or traffic.

Not all is lost for fatigue-critical work. Techniques such as toe grinding or TIG dressing can reduce the stress concentration at the fillet toe and improve fatigue performance. These post-weld treatments smooth the transition and are often specified for bridges, cranes and other high-cycle applications. However, they are not a substitute for correct sizing, and they must be performed with care to avoid thinning the base metal.

Quick Note: A weld that looks perfect to the naked eye can still be a fatigue risk. Visual acceptance alone will not catch a poor toe profile or an undersized throat.

Load Paths, Eccentricity, and Governing Codes

Geometry can also rob a fillet of its assumed capacity. Single-sided fillets introduce eccentricity, creating a moment in addition to shear and placing uneven stress on the connection. Poor load paths, such as a bracket pulling off a thin flange, can induce prying forces that far exceed the loads the engineer intended. These are the exact scenarios where craftsmanship must align with design, a principle that is drilled into those building real welding skills and into large-scale fabrication alike.

These requirements are not universal. AWS D1.1 governs structural steel, the ASME BPVC governs pressure equipment, and AISC 360 governs building steel design — different codes, sometimes with different allowable stresses, effective throat rules or fatigue details. Workplace safety also remains key, with OSHA and NIOSH outlining general welding controls.

Consider a bracket that “passed” a visual inspection but cracked in service because the design credited a full-strength fillet that the joint geometry never could deliver. With no account for eccentricity, a subpar toe, or cyclic loads, the throat that was assumed simply was not there. It is a stark reminder: for a fillet weld, the angle, the axis, and the cycles will always matter just as much as the leg size.

Common Fillet Weld Defects and Mistakes: Undersize, Undercut, Overlap, Incomplete Fusion, and Excess Convexity

A fillet weld can look flawless and still be worthless. A smooth cover pass hides everything beneath it, and the defects that sink fabrications are usually invisible. Here is what goes wrong, why, how inspectors catch it, and the fix.

Undersize Fillets: The Most-Cited Code Violation

An undersized fillet is one of the most frequently cited AWS D1.1 code violations in fabrication. Welders underestimate the leg, hold too much travel angle, or lose filler at the toes. Minimum fillet sizes are governed by the thickness of the thicker part, and a weld that measures small is rejectable no matter how pretty it looks. Inspectors verify size with a fillet weld gauge. Turn up the current slightly, shorten the arc, square your work angle, and slow down at the toes. Practice fixes size; see our guide to building real welding skills.

Undercut: Melting an Edge You Never Fill

Undercut is a groove melted into the base metal along the toe and left unfilled — usually from too much current, a long arc, or an angle that points heat at the plate edge. It concentrates stress right where fatigue cracks start. Visual inspection catches obvious undercut; magnetic particle testing finds the finer ones. Reduce amperage, tighten arc length, and pause at each toe so the puddle washes in.

Overlap and Cold Lap: Hidden by a Nice-Looking Bead

Overlap is filler metal that rolls onto the base metal without fusing. Run too cold or drag too fast and the molten metal rides over the surface, leaving a mechanical ledge instead of a bond. Cold lap is the dangerous version because it looks like a normal toe from outside — visual inspection cannot find it, but magnetic particle or ultrasonic testing can. The fix is heat and patience: raise current, slow down, and let the arc melt the base edge before filler flows over. Our overlap welding defect guide covers detection and repair in depth.

Incomplete Fusion and Lack of Root Penetration

This is the silent killer. Root gaps fit too tight, travel too fast, insufficient current, or the wrong polarity means the arc never reaches the root, so the weld sits on unfused metal. Finding it typically requires volumetric or surface NDT such as ultrasonic, magnetic particle, or dye penetrant testing. Check polarity before you blame technique; swapping leads fixes more of these than welders expect. Our guide to TIG welding polarity explains how DCEN and DCEP move the heat.

Excess Convexity, Concavity, Craters, and Spatter

A tall, cold bead with a sharp toe is excessive convexity — too little heat for the wire feed, or travel too slow. It creates a stress riser that looks like strength and behaves like weakness. Excessive concavity comes from over-grinding or too hot a puddle and thins the throat. Crater cracks form when you break the arc and leave a shrinking puddle; add a small fill before you stop. Spatter and ragged restarts usually mean a loose contact tip or poor gas coverage.

The 1/4-Inch Fillet That Passed Visual and Failed UT

One shop shipped bracket assemblies with quarter-inch fillets that looked textbook, and the visual inspector signed off. Then ultrasonic testing flagged the root of nearly every weld: the cover pass was beautiful, but underneath, the root had never fused — a classic cold lap. Fit-up was tight, travel was fast, and the operator had learned pretty beads without learning sound ones. Rework cost more than the original job.

Quick Note: Buy a fillet weld gauge and use it on every critical joint. The eye consistently overestimates leg size — a weld that looks like 1/4 inch often measures 3/16. Measuring takes seconds; failing a code inspection costs days.

Most of these defects share two causes: too little heat driven into the joint, and too little patience at the toes. Run coupons at your intended settings, verify size with a gauge, and treat every cover pass as a place where problems hide rather than disappear.

Choosing and Making the Right Fillet Weld: Process Selection, Technique, Symbols, Inspection, and Key Takeaways

All the theory — profiles, joint types, continuity, legs and throats — lands in one place: the arc you actually strike. Here is how to pick a process, run the bead, and prove it passes before it ever leaves your bench.

Match the Process to the Job

SMAW (stick): outdoor, dirty, rusty, and repair work. Electrode selection is simple and forgiving.
GMAW: speed and thin material in controlled shop conditions. See our MIG welding at home guide.
FCAW / gasless flux-cored: high deposition and windy field conditions; a self-shielded flux shrugs off wind that blows away solid-wire shielding gas. Read gasless MIG welding and polarity for flux core welding.
GTAW: high-quality small fillets, stainless, thin sections.
SAW: heavy plate, flat or horizontal, enormous deposition.

Technique Essentials

Work angle first: for an equal-leg T-joint fillet, hold roughly 45° to both members, then bias toward the thicker member or heavier leg when the legs are unequal. Travel angle next: gas-shielded GMAW generally pushes at about 5–15°, while SMAW is typically dragged. Beyond angles, keep travel speed consistent, set amperage to the material thickness, confirm polarity for the process, and tie in every restart so you never leave a cold lap behind. On vertical-up fillets, drop the amperage a touch and step the puddle in a small triangle so it doesn’t sag ahead of you.

Electrode selection has its own plumbing. A 1/8 in E7018 on a typical horizontal fillet runs and behaves differently from vertical-up work, where amperage, arc length, and manipulation all change — our 3/16 E7018 amperage guide walks through that same logic at a heavier size.

Sequencing matters as much as the bead. Balance welds around the neutral axis, alternate sides, backstep long runs, and clamp or fixture parts to pull distortion the other way. Stagger your stops between layers rather than stacking them in one column.

Quick Note: The drawing’s fillet symbol tells you the leg size, not the throat. If you gauge only what looks “about right,” you can finish 20–30% under size and still pass a casual glance.

Inspection and Acceptance

Inspect in order: verify size with a fillet gauge against the symbol, then check undercut, convexity, overlap, and incomplete fusion at the toes and root. Visual inspection catches the majority of fillet defects before they become expensive repairs. Where the code specifies NDT (MT, PT, UT, or RT), run it, log the results, document any repair, and re-inspect after rework. Log the leg size you measured and the gauge you used, not just a pass/fail tick. Most project specs trace their acceptance criteria to AWS standards, with structural rules from AISC.

Fume, PPE, and the People Doing the Work

Stainless is the process that catches shops out: its fume can contain hexavalent chromium, which OSHA regulates at a permissible exposure limit of 5 µg/m³ as an 8-hour time-weighted average under 29 CFR 1910.1026. Fume extraction, local exhaust, and PAPR hoods are the answer, not willpower. NIOSH keeps a broad welding safety resource covering fumes, radiation, and PPE, and OSHA’s 1910.252 sets the baseline for ventilation and protection. A well-planned bay makes this easy — see our guide to planning a safe welding shop. Don’t skip the small stuff either: proper protective eyewear for welding is part of the kit.

Key Takeaways

Fillet welds are classified by profile, joint type, continuity, and side count.
Size follows leg and throat logic: the leg is what you draw, the effective throat is what carries.
Strength follows the effective throat and load angle.
Fatigue and defects — not static strength alone — decide real service life.

So pick the right type, measure it rather than eyeball it, and weld it the same way every time. Consistency, not heroics, is what makes a fillet weld trustworthy. Do that, and the next weld — and the one after it — will look and test the same.

Frequently Asked Questions

What is the difference between a fillet weld and a groove weld?

A fillet weld is a roughly triangular weld that joins two surfaces meeting at approximately right angles, typically at a T, lap, or corner joint. A groove weld fills a gap or prepared bevel between two edges, such as a butt joint, and can achieve full penetration through the joint thickness.

How many types of fillet weld are there?

There is no single number, because fillet welds are classified several ways: by profile (convex, concave, flat/mitre), by joint type (T, lap, corner, edge), by continuity (continuous, intermittent chain, intermittent staggered), by side count (single-sided, double-sided), and by process (stick, MIG, flux-cored, TIG, submerged arc).

What is the difference between leg size and throat thickness on a fillet weld?

Leg size is the length of one of the two fused legs forming the triangle; it is usually the dimension shown on a drawing. Throat thickness is the shortest distance from the root to the face of the weld, and for an equal-leg 90-degree fillet the theoretical throat is 0.707 times the leg size. Strength calculations use the effective throat, not the leg.

Is a convex or concave fillet weld stronger?

For the same measured leg size, a convex fillet can have slightly more throat because the face bulges outward, but it has a sharper toe notch that reduces fatigue life. A concave fillet has a smoother toe and better fatigue performance, but its effective throat is reduced by the depth of concavity — so over-grinding a concave fillet makes it undersized.

What is the minimum fillet weld size?

AWS D1.1 bases the minimum fillet weld size on the thickness of the thinner part: commonly 1/8 in for material up to 1/4 in thick, 3/16 in over 1/4 to 1/2 in, 1/4 in over 1/2 to 3/4 in, and 5/16 in for thicker parts. The minimum exists to ensure enough heat input for proper fusion and to avoid cold, crack-prone beads.

What causes an undersized fillet weld?

An undersized fillet usually results from a leg dimension that is smaller than specified, which can come from incorrect sizing on the drawing, travel speed too fast, incorrect travel or work angle, insufficient filler deposition, or over-grinding after welding. It is one of the most common code violations, which is why measuring with a fillet weld gauge matters.

What is an intermittent fillet weld and when should I use one?

An intermittent fillet weld is applied in segments with gaps between them, specified as length and pitch (such as 3 in of weld on 9 in centers). It is used to reduce weld volume, cost, and distortion where the joint does not require continuous load transfer. It should not be used where the joint must be watertight or where fatigue or corrosion resistance governs.

Can a fillet weld be as strong as a groove weld?

A fillet weld can be extremely strong in shear but typically does not develop the full tensile strength of the base metal across the joint the way a full-penetration groove weld does. Where a connection must transfer full member strength, especially in tension or under fatigue, engineers usually specify a groove weld or a combination of groove and fillet welds.

Does it matter which process I use for a fillet weld?

Yes. Stick and flux-cored welding tolerate wind, dirt, and outdoor conditions and deposit quickly; MIG is fast and clean for shop work and thin material; TIG produces high-quality small fillets; and submerged arc is used for heavy, high-deposition plate work. The process affects penetration, appearance, spatter, distortion, and what code or procedure may apply.

How do I inspect a fillet weld?

Start with visual inspection: check leg size with a fillet weld gauge, look for undercut, overlap, excessive convexity or concavity, cracks, and crater defects, and verify length and pitch. Where the specification requires it, follow with magnetic particle, dye penetrant, ultrasonic, or radiographic testing to find incomplete fusion and lack of penetration that eyes cannot see.