Best CA Glue for Woodworking: What the One Real Test Found, and What Nobody Measured

My picks

What I own vs. what I researched

Every score here comes from the two measured studies and the forum threads linked below, not from my bench. When I say which bottle to buy, that's me reading the load-cell results, not me squeezing the bottle.

How this guide decides

Almost nobody has put CA glue on wood and measured the break. So this page says, claim by claim, which ones rest on a load cell and which rest on woodworkers arguing in a forum.

• James Wright's "Great Glue Test" (Wood by Wright). The only large measured study I found that puts CA on wood. 32 glues, each broken 10 times across four scenarios: long grain to long grain, long grain to end grain, gap filling, and exterior exposure. Measured on a calibrated load cell, not by hanging weights. His CA entries were generic super glue, Fix All, and FastCap 2P-10 in thin, thick and gel, each also run with activator. One caveat I won't paper over: his write-up reports rankings and directional findings, not per-glue break figures in pounds. So this page cites his conclusions and does not invent numbers for them.
• Matthias Wandel's joint-strength testing (woodgears.ca). Measured break loads on scarf, butt and cross-grain joints. He tested no CA glue at all, which is exactly why he's here: his numbers set the PVA baseline that CA has to be judged against.
• Project Farm's super-glue comparison (2018). Six brands, lateral, tensile and torsional tests, on metal and plastic. Wood was not part of it, and by the account of contemporaneous coverage he declined to crown an all-around winner. His videos carry his own standing disclaimer that they're "only for entertainment purposes." I cite him for what he tested and nothing more.
• Manufacturer technical statements. BSI's own "Ask the Glue Pros" and Starbond's storage guidance, used only for checkable specs (shelf life, storage, what CA won't bond to), never for performance claims.
• Where the woodworking forums agree. Sawmill Creek, the American Association of Woodturners and Woodworking Talk, where the people who go through CA by the gallon argue about it.

The finding that contradicts the bottle: thicker is stronger

The second half of that result is the one that should change what's on your shelf. In his long-grain-to-end-grain, gap-filling and exterior scenarios, the CA glues did extremely well. In a number of cases they beat standard epoxies and PVAs outright. End grain and gap filling are exactly the two jobs where woodworkers reach for epoxy out of habit, and this is real evidence that a thick CA competes there.

Thin CA is for different work entirely: wicking into a hairline split, stabilizing soft or punky wood, running into a crack you can't open up. That's a penetration job, not a strength job. Buying thin and expecting it to hold a joint is the exact mistake the viscosity finding corrects. The thin bottle goes where the thick one can't reach. It doesn't go where the clamp should be.

Does the accelerator weaken the bond?

The manufacturers don't entirely disagree. They just frame it as a difference between accelerators. BSI's own position is that many accelerators "create excessive heat that may cause the CA to become very brittle." BSI also says theirs "has the least affect on the strength of CA." That's a marketing claim, and a useful one. It concedes the point: the question isn't whether accelerator costs you strength, only how much. Starbond's selling point is similar in spirit. Theirs won't whiten a glue-up on over-spray the way others do.

The practical version: on a jig or a miter where speed is the whole point, spray away. On a repair you want to hold, mist the accelerator on the opposing surface instead of into the glue line, and give it time.

Is CA glue strong enough for actual joinery?

Wandel's measured joint tests are the cleanest support for that. Across spruce and maple, well-made PVA joints broke in the range of a few hundred pounds. The failures were often in the wood, not the glue line. Titebond III topped his maple scarf joints at 350 lbs. His summary judgment is the one to carry into the shop: for most applications, any good-quality wood glue will be stronger than the wood itself. There is no strength headroom left for CA to win back.

Where CA earns its shelf space is the work PVA can't do. Holding a part while the real glue cures. Fixing a chipped edge. Gluing a jig you'll destroy later. Filling a void. Sticking a turning blank to a waste block. Stabilizing a crack in a workpiece that's already on the lathe. That's a long list. It's just not joinery.

Every test, one table

Every source is linked under "How this guide decides" above. Wright published rankings and directional findings, not per-glue break loads, so no pound figure is pinned to any product here. Where I have no number, I say so instead of inventing one. Project Farm's figures are not used for any wood claim, because he did not test wood.

Why your last bottle died before you finished it

The manufacturers are unusually candid here. BSI names the mechanism directly: heat and moisture are the two main factors that shorten CA's life. BSI puts its unopened figure at over two years at 74 °F, and past four years in a sealed container with desiccant in the fridge. Starbond guarantees 20 months and recommends storage below 40 °F.

Three caveats that get lost in the "keep it in the fridge" advice repeated everywhere:

• Don't refrigerate the bottle you're actively using. BSI is explicit: a frequently opened bottle shouldn't go in the fridge, because condensation forms inside. You'd be adding the exact contaminant that kills CA. Cold storage is for stock, not for the working bottle.
• Cold glue cures slower. Starbond notes refrigeration doesn't change the consistency, but a cold bottle takes longer to set. Let it come up to room temperature for a minute or two first.
• Never store accelerator with the CA. Both makers say it. It's the fastest way to ruin a whole shelf.

The clog itself is mechanical, not chemical. Wipe the nozzle, tap the bottle on the bench so the glue in the tip draws back down, then cap it. That habit outlasts any brand choice.

Common questions

Thin, medium or thick: which CA should I buy first?

Medium, if you're buying one. It handles most repairs and small assemblies without running off the workpiece. Buy thin second, specifically for cracks. It's a penetration tool, and the measured testing found it's the weakest tier, so don't ask it to hold anything structural. Thick and gel are the strongest on wood and the least controllable.

Is CA glue waterproof?

It held up better than expected. The CA glues performed well in Wright's exterior-exposure scenario, in some cases beating PVAs and epoxies. That's a stronger result than CA's reputation suggests, but it's one study, and it isn't a substitute for a glue rated for outdoor structural use.

Does CA glue work on oily woods like teak or rosewood?

Less reliably, and the mechanism is worth knowing. CA cures on contact with trace surface moisture, so anything that changes the surface chemistry changes the result. Wiping the joint with a solvent right before gluing is the standard workaround. One hard limit: nothing made from or coated with PTFE will bond to CA at all. That's BSI's own stated limit.

Why do woodturners use so much of it?

Because on a lathe you can't clamp anything. CA fills voids and stabilizes cracks in a spinning workpiece, glues blanks to waste blocks, and doubles as a finish on small work. It's the clearest case of CA being used for what it's good at, speed and access, rather than for strength.