Kerf Allowance: Why Your Laser Cut Joints Don't Fit

A laser does not cut along a line. It burns a channel, and that channel has width. That width is the kerf, and it is the single most common reason a box that looks perfect on screen either falls apart in your hands or refuses to go together at all.

What kerf actually costs you

On 3 mm plywood with a typical CO₂ laser, kerf runs somewhere between 0.15 mm and 0.25 mm. That sounds like nothing. It is not nothing.

Cut a finger joint with five fingers per edge and you have ten cut lines, each removing roughly 0.2 mm. Across that edge you have lost 2 mm — most of a material thickness. The fingers rattle in their slots, glue does the structural work it was never meant to do, and the box racks out of square the first time someone picks it up.

Go the other way — design a slot exactly the width of your material — and the kerf removes material from the slot as well, making it wider than intended. Now the joint is loose in a different direction. Neither mistake is visible until the parts are off the bed.

Measure it once, properly

Do not trust a number from a forum post. Kerf changes with material, thickness, power, speed and even how clean your lens is. Measure your own:

1. Cut a 50 mm square from the material you actually plan to use.
2. Measure the square with calipers. It will come out slightly under 50 mm.
3. Subtract: kerf = 50 − measured. Then halve it, because the beam removed material from both sides of the path.

A cleaner variation: cut ten 10 mm strips side by side from one sheet, measure the total width of all ten together, and divide the shortfall by ten. Averaging across ten cuts cancels out most of your measurement error.

Designing with the number

Once you know your kerf, apply it as an offset rather than redrawing geometry by hand:

• Slots and holes get smaller by one kerf so the finished opening is correct.
• Tabs and fingers get larger by one kerf so the finished part is correct.
• Outside profiles move outward by half a kerf if the finished dimension matters.

In practice this means every joint in a file is drawn at nominal size, then offset in the direction that compensates for the burn. Most vector tools do this with a single offset-path operation on a selected set of contours.

Press fit, slip fit, or glue

Not every joint wants the same tolerance, and this is where a lot of otherwise careful files go wrong.

A press fit — a lid that stays on by friction — wants roughly half a kerf of interference. It should need a firm push. A slip fit — a drawer, a sliding tray, anything that moves — wants a quarter to a half millimetre of clearance on top of kerf compensation. A glued joint wants a light slip fit, because a joint hammered together squeezes out the adhesive and ends up weaker than one that seats easily.

Deciding which of the three you want before drawing the joint is most of the work.

The test cut is not optional

Every design I release is cut before it ships, and the first cut is almost never the last. What you are checking is narrow and specific: do the joints seat with the right amount of resistance, does the assembly square up, and does the engrave layer read as engrave rather than cut.

Ten minutes on a scrap sheet costs less than a customer telling you their box does not close.

Related guides: How to design a laser cut box with finger joints · Resizing files for a different material thickness · Tab and slot desktop organizer

---

Working on something that has to fit? I design production-ready laser cut files with kerf compensation measured for your machine and material. Send me the brief.