How Strong Is Wood Glue? What Actually Fails in a Joint

How strong is wood glue? On a properly made long-grain joint, stronger than the wood it holds together. That is not a marketing line, it is what the published shear numbers show, and it is the reason a glue joint that fails usually fails by tearing wood fibers rather than by letting go along the glue line.

I learned the difference the irritating way. A walnut stool I built years ago came apart at the leg seams, and my first assumption was that the glue was too weak for the job. It was not. I had grabbed an unlabeled craft PVA off the shelf and spread it on surfaces I had sanded smooth — two separate mistakes, neither of which had anything to do with how strong wood glue can actually be. I rebuilt the same joints with a Type I glue on freshly jointed faces and that stool is still in one piece.

What follows is not a lab report. I do not own a load cell or a test frame, and anyone who tells you they measured glue strength to the pound in a two-car garage is estimating. What I can give you is the strength data the manufacturers actually publish, the one failure mode that tells you whether a joint is sound, and the handful of mistakes I have made often enough to spot from across the shop.

Glue	ANSI/HPVA rating	Water resistance	How it fails in my shop
Unbranded craft PVA	Unclassified	Effectively none	Separates along the glue line; creeps under sustained load
Titebond Original	Type II	Water resistant, not waterproof	Wood tears first indoors; the glue line suffers in a humid garage
Titebond II Premium	Type II	Water resistant	Wood tears first; not for standing water
Titebond III Ultimate	Type I	Waterproof	Wood tears first

What 3,600 psi actually tells you

Franklin publishes a bond strength figure for Titebond Original of 3,600 psi on hard maple, tested to ASTM D-905, with 77 percent wood failure at room temperature. That second number is the one worth staring at. “77 percent wood failure” means that across the test blocks, most of the fracture ran through maple rather than through adhesive. The glue was not the weak link; the wood was.

Hold that against what furniture actually has to survive. A chair leg or a stool rail is nowhere near 3,600 psi in service, and the cross-grain strength of common furniture hardwood is well under the shear strength of a sound PVA glue line. Which is why the honest answer to “is glue alone strong enough” is almost always yes, provided the joint has long-grain surfaces in contact and you did not sabotage the prep. Current figures for any specific product belong on the manufacturer’s technical data sheets rather than in a blog post, because formulations change.

The only glue test that matters in a home shop

You cannot measure pounds of force without a frame, but you can read a broken joint, and that tells you nearly everything. Glue two offcuts of the same stock you are about to use, let them cure, then put the test piece in a vise and break it. Then look at the fracture.

If wood fibers tore — a fuzzy, splintered surface on both halves — the adhesive outperformed the wood and the joint was sound. If the two faces came apart clean, with a glossy or patchy film and no torn fiber, the glue line was the limiting factor, and that points at the glue, the surface prep, the clamping pressure, or stressing the joint before it cured. Clean separation is the only result that should worry you. In my experience the cheap unbranded PVAs are the ones that come apart clean, and every name-brand yellow glue I have broken has taken wood with it.

Why butt joints still fail, and it is not the glue

End grain is a bundle of open straws. It wicks adhesive away from the joint line and leaves you a starved, shallow bond over a tiny effective area, so a butt joint is weak no matter which bottle you reach for. Changing glue does not change the geometry. If a butt joint has to carry load, add mechanical help — a dowel, a spline, a floating tenon, a pocket screw — or size up to a joint with long-grain contact. This is the one place where people blame the adhesive for a problem they designed in.

Edge joints and the creep problem

A glued-up panel lives under constant low-level stress from seasonal movement, and that is a different demand than a single hard pull. What matters is creep: whether the glue line slowly deforms under sustained load instead of holding its shape. Cheap craft PVA is the worst offender I have used, and it shows up not as a dramatic failure but as a panel seam that has gone slightly proud or opened a hair two winters later. A panel destined for a humid garage or any exterior exposure is worth a Type I glue, which is most of the reason I stopped using Titebond Original on shop fixtures.

Mortise and tenon: the glue is not doing the heavy lifting

A well-fitted mortise and tenon is a mechanical joint. The shoulders resist racking and the cheeks carry the load through long-grain-to-long-grain contact; the adhesive mostly keeps it from working loose over years of use. That is why glue choice matters least on the strongest joint in the furniture and most on the marginal ones. The same logic covers loose-tenon systems like a Domino — a machined floating tenon is still long grain against long grain, so any decent Type I or Type II PVA does the job.

Type I vs Type II: what the ratings actually promise

Wood glues sold in the US are commonly described against the ANSI/HPVA HP-1 standard for hardwood plywood, informally split into Type I (the more demanding water-resistance specification), Type II (moisture resistant), and unclassified. Titebond III is published as meeting ANSI/HPVA Type I; Titebond Original and Titebond II carry Type II. Unbranded craft PVA carries nothing, which is itself the useful signal.

For interior furniture in a climate-controlled house, Type II is genuinely fine and I have plenty of it in service. My garage runs humid through the summer, so for anything that lives out there — benches, jigs, shop fixtures, outdoor pieces — I use a Type I glue and stop thinking about it.

One claim worth handling carefully: Titebond III is widely described as FDA approved for indirect food contact, which is the line people cite when building cutting boards. The practical reality is that the FDA does not “approve” adhesives. That is the manufacturer’s own compliance claim about an indirect food-contact application once the glue is fully cured. For a board on your own counter it is a non-issue. If you are selling boards and need it in writing, ask Franklin for their letter of guaranty rather than trusting a forum post or, for that matter, this one.

Surface prep ruins more joints than glue choice

Do not sand your glue surfaces. Sandpaper burnishes and clogs the open fiber the adhesive needs to key into, and it rounds the edges of the joint so the faces no longer mate. Mating surfaces off a jointer, a table saw blade in good shape, or a sharp hand plane give you flat, freshly opened fiber, which is exactly what you want. This single habit fixed more of my failures than switching brands ever did.

Two more that cost me joints early on. Clamping pressure is a real variable, and Titebond publishes guidance by wood density — roughly 100–150 psi for softwoods, 125–175 for medium woods and 175–250 for hardwoods — so a couple of light spring clamps on an oak glue-up is not enough. And cure time is not clamp time: Titebond’s documentation puts full strength at around 24 hours, so a joint that feels solid when you pull the clamps at thirty minutes has nowhere near its final strength. Stress it that afternoon and you will get a clean separation and blame the glue.

What I keep on the shelf

Titebond III is my default, and the reason is boring: it is Type I, so I do not have to think about where the finished piece will live. That is worth more to me than any strength difference between the name-brand yellow glues, which in practice all exceed the wood anyway.

It is not free of annoyances, though:

• Squeeze-out is visible on dark woods like walnut, and it needs cleaning up properly rather than smearing it around with a wet rag.
• A 16 oz bottle is a lot of glue. If you build one project a year it will likely skin over before you reach the bottom, and a smaller bottle of Titebond Original is the more sensible buy for interior work.
• It is the most expensive of the three, for a water-resistance rating that interior furniture does not need.

So: if everything you build stays indoors, Titebond Original at 3,600 psi on hard maple is more glue than your furniture will ever ask for. If your shop is damp, or the piece goes outside, or you simply do not want to think about it, pay for the Type I. Either way, the glue is not what is going to fail — the prep is.

Frequently Asked Questions

How strong is wood glue compared to the wood itself?

On a properly made long-grain joint, stronger than the wood. Franklin publishes a bond strength of 3,600 psi on hard maple for Titebond Original under ASTM D-905, with 77 percent wood failure, meaning most of the fracture ran through the maple rather than the adhesive. That is why a sound glue joint breaks by tearing wood fibers.

Why did my wood glue joint fail?

Look at the broken faces. If they separated clean, with no torn fiber, the glue line was the limiting factor, which points at a cheap unclassified PVA, burnished or starved surfaces, too little clamping pressure, or load applied before the glue cured. If wood fibers tore, the joint was sound and the problem is the joint geometry or the load, not the adhesive.

Should I sand surfaces before gluing?

No. Sanding burnishes and clogs the open fiber the adhesive needs to penetrate, and it rounds the joint edges so the faces stop mating. Prepare glue surfaces with a jointer, a clean saw cut, or a sharp hand plane instead.

Is Titebond Original waterproof enough for a garage?

No. Titebond Original is published as ANSI/HPVA Type II, which is moisture resistant rather than waterproof. For a humid garage, shop fixtures or anything with exterior exposure, use a Type I glue such as Titebond III.