Fit data

The Clearance Index

Every guide says to leave 0.2 to 0.3 mm for a 3D-printed fit. SpeakCAD's users print a five-minute test of the part that has to fit and report what happened. This is what they actually got.
187first triesOne person, one model, one fit. A coupon printed again after a verdict does not count twice.
44%fitted on the first print83 perfect, 54 too tight, 50 too loose.
42%fitted at 0.3 mm22 of 52 first tries at the usual 0.3 mm.

What should your fit leave?

Pick the kind of fit. The range comes from the clearances people reported perfect on the first try: the middle half of those reports, not a rule of thumb.
0.20 mmmedian clearance reported perfect
0.15 mm – 0.35 mm middle half of the reports43% fitted on the first try (15 of 35)15 perfect reports behind the range
A magnet, an insert, a phone: something sits inside a printed pocket and should not rattle or stick.

Get the report by e-mail

The figures for your kind of fit and your printer, each with how many first tries are behind it, plus the free coupon to find your own number. One e-mail.
No account, no card. Every e-mail carries an unsubscribe link.

By kind of fit

A pocket does not behave like a slot, and a thread does not behave like a lid. The first try is counted once per person, model and fit.
Kind of fitFirst triesFittedToo tightToo looseFirst-try fitMedian perfectMiddle half
Pockets and cradles351514643%0.20 mm0.15 mm – 0.35 mm
Bores, pins and sockets33179752%0.20 mm0.00 mm – 0.30 mm
Slots and slides28781325%0.30 mm0.23 mm – 0.35 mm
Lids, caps and snaps24147358%0.25 mm0.20 mm – 0.30 mm
Threads and bayonets1996447%0.40 mm0.30 mm – 0.60 mm
Other fits4821101744%0.30 mm0.20 mm – 0.40 mm

By clearance

The same clearance, different results: this is why one fixed number cannot be right for every printer and every fit.
ClearanceFirst triesFittedToo tightToo looseFirst-try fit
0.00 mm22116550%
0.05 mm10100%
0.10 mm530260%
0.15 mm961267%
0.20 mm33178852%
0.25 mm1026220%
0.30 mm5222161442%
0.35 mm312033%
0.40 mm27128744%
0.45 mm210150%
0.50 mm1533920%
0.60 mm532060%
0.75 mm1100100%
1.00 mm10100%
2.00 mm1100100%

By printer

The printers with at least 10 first tries. Each one has its own page with the same figures.
PrinterFirst triesFittedFirst-try fitMedian perfect
Bambu Lab P2S 0.4 nozzle12433%0.20 mm
Pages appear at 10 first tries on the same machine. The index currently names 1 printer.

How the index is built

The data is self-reported, small, and honest about both.
  1. A fit test is a small coupon cut from the part that has to mate with a real object: a screw hole, a bearing seat, a lid lip, a phone slot. It prints in about five minutes and uses a gram or two of filament.
  2. The maker tries the real object in it and reports too tight, perfect or too loose. The verdict calibrates the printer on the account (lib/fit.ts), so the next model is compiled with the clearance that printer needs.
  3. Only the first try per person, model and fit counts. A coupon printed again after a verdict confirms the fit; it is not a second opinion, so it does not weigh more.
  4. The results are self-reported, and most tests do not record the filament or the printer. The figures below are a map of what works, not a guarantee for a given spool.
  5. SpeakCAD's index grows every day. The page is rebuilt hourly; the date under the figures is the newest verdict it saw.
Newest verdict: · page reviewed 2026-11-22

Clearance questions

There is no single number. Across 187 first tries in the Clearance Index, 83 fitted, 54 were too tight and 50 too loose: 44% fitted on the first print. At the usual 0.3 mm, 22 of 52 first tries fitted (42%), 16 were too tight and 14 too loose. Start from the range for the kind of fit, then print a five-minute test coupon on your own printer.

At the usual 0.3 mm, 22 of 52 first tries fitted (42%), 16 were too tight and 14 too loose. It is a good starting point and a poor guarantee: the same 0.3 mm can be tight on one printer and loose on another, and a pocket and a slot do not behave the same way. The Clearance Index lets you pick the kind of fit and see what people reported perfect.

The printer, the filament, the flow and first-layer settings, and the shape of the fit all move the result. A pocket shrinks inward and its corners round; a slot tends to run loose. That is why SpeakCAD calibrates each printer from a printed coupon instead of applying one global offset.

187 first tries with a verdict, from SpeakCAD accounts that printed a Fit Check coupon or judged a full part. A test is counted once per person, model and fit; a coupon printed again after a verdict does not add a second data point. The index is rebuilt hourly.

Yes. Make a part in SpeakCAD, open Fit Check, print the coupon and report how it fitted. The verdict calibrates your printer and, with your account's other tests, feeds the same public figures. The coupon itself is free.

Print the coupon, then print the part

Fit Check cuts a five-minute test from the part that has to fit. Report how it went and every model after it is built with your printer's clearance.