How to Calculate Bicycle Spoke Length: Hub Dimensions, ERD & Wheel Dish

The method used in a Japanese wheel-building workshop with nearly two decades of builds — with links to the original posts so you can check the source.

Spoke length is the single most consequential number in a wheel build. Too short, and the nipple can't grip enough thread — the joint is weak. Too long, and the spoke bottoms out, or pokes past the nipple into the rim bed and becomes a puncture risk. This guide distills the method one workshop has refined over thousands of builds. Every figure below comes from a specific build write-up, linked at the end of each section so you can verify it yourself.

1. The three measurements you actually need

To calculate spoke length you need exactly three numbers:

1. The rim's effective measurement — its ERD, derived from the rim's support length (the diameter at which the nipple seats).
2. Hub flange width — the distance from the outer face of one flange to the outer face of the other.
3. Hub flange diameter — the pitch circle the spoke holes sit on.

Feed these into any reputable spoke calculator, per side, and it returns a length. The whole craft is in measuring them correctly — a quality digital caliper is the one tool that makes this reliable — and in understanding two subtleties: how a rear hub's two sides differ, and what the calculated number actually represents (covered below).

Source: Nomu Lab #227 — On wheel dish

2. Hub flange width and wheel dish

On a front rim-brake wheel or a track hub, the two flanges sit symmetrically about the centre. A rear wheel is different: the cassette forces the drive-side (right) flange inward, so the left flange ends up farther from the centre than the right. That asymmetry is the dish (Japanese: ochoko).

Call the hub's total flange width W, the left (non-drive) flange's distance from centre L, and the right (drive) flange's distance R. On a dished rear wheel L > R, and the true dish is L − R.

Here is a trap that catches many builders. Shimano's catalogs (since 2012) don't print L and R directly — they print flange width and a "dish amount" that is actually (L − R) / 2, i.e. half the real dish. To recover the numbers you need from a Shimano spec:

• half = W ÷ 2
• L = half + (catalog dish)
• R = half − (catalog dish)

Source: Nomu Lab #227 — On wheel dish

3. Why dish matters more than the spec sheet suggests

A bigger dish increases the left/right spoke-tension imbalance: the drive side is tensioned hard while the non-drive side runs slack. But tension imbalance is recoverable — you can claw it back with lacing choices (different spoke counts or gauges per side). What you cannot fix by lacing is a hub whose left flange doesn't protrude far enough: narrow flange spacing gives a laterally flexible wheel, full stop.

So the workshop's rule for judging a rear hub is to look at how far the left flange protrudes, not just the headline flange width. And beware spec-sheet games: Shimano's move to 11-speed narrowed the freehub-side flange by about 2 mm, which inflates the apparent dish number while actually reducing total flange width — worse for stiffness, not better.

Source: Nomu Lab #227 — On wheel dish

4. ERD: don't trust the chart

The effective rim diameter has the biggest effect on the result, and it's also the number most often gotten wrong — because manufacturers define and publish it inconsistently. The workshop's rule: measure your own rim, and for critical builds, verify by test-building.

A real example: when a customer needed to order carbon spokes — which, unlike steel, can't be re-cut if wrong — the workshop test-built the wheel with sacrificial spokes rather than trust the rim maker's chart. The maker's ERD definition was idiosyncratic and, even after accounting for that, still off. The discrepancy wasn't a rounding question of whether 285.5 becomes 285 or 286; it was large enough to make the wheel unbuildable on one side.

If you can re-cut steel spokes, an error is recoverable. If you're committing to a fixed length (carbon, or a bulk order), measure twice — ideally with a test build.

Source: Nomu Lab #6134 — Determining spoke length for certain

5. What a correctly-sized spoke looks like

The workshop's standard is simple: the spoke should finish flush with the end face of the nipple, or barely showing. If you must err, err long — one thread too long is better (or at least less bad) than one thread too short. Two or more threads short is a sign of sloppiness or a calculation error.

Counter-examples abound on factory wheels. Some Campagnolo wheels show spokes poking three or more threads past the nipple, which looks wrong but isn't — those spokes simply have longer-than-standard thread, so the calculation is fine. Other factory wheels genuinely are miscalculated: some disc-brake fronts use the same length on both sides of a hub that needs two different lengths.

On rounding, the workshop rounds at the half-millimetre (285.49 → 285, 285.50 → 286), but because a 0.01 mm difference shouldn't swing the result a whole millimetre, anything in the 285.4–285.6 band is often just cut at 285.5 — possible because they cut every spoke from 315 mm plain-gauge stock, giving any length in 0.5 mm steps.

Source: Nomu Lab #7000 — On spoke length, #6134

6. A subtlety: inpoke vs outpoke length

The length your calculator returns, from flange-outer-face measurements, is strictly the inpoke length — the spoke laced from inside the flange outward. An outpoke spoke (laced outside-in) should, in principle, be calculated to the inner face of the flange. In practice the difference is small enough to be absorbed by the nipple's thread length, so you don't need to calculate two lengths per flange.

Source: Nomu Lab #7000 — On spoke length · see also the glossary entry for inpoke / outpoke