Park Tool sets the convention in its truing guide: spokes on the same side of a wheel shouldn't spread more than ±20 % from that side's average tension. There's no published test behind the figure — it's a shop criterion, not a standard. The generic reference range that goes with it, 100 to 120 kgf, is given with the tire unpressurized, a condition almost everyone drops when quoting it.
It's the number that decides, spoke by spoke, whether a wheel is done or the wrench keeps turning.
Park Tool publishes it in its truing guide, «Spoke Tension Measurement and Adjustment», dated 2021-04-06: spokes on the same side of the wheel should sit within ±20 % of that side's average tension. It's the figure repeated at every truing stand.
There's no published test behind it. We went looking and it isn't there, not in the guide itself, not cited in any DT Swiss document. It's a shop criterion, set by the maker of the tool used to measure it, not a standard verified in a lab. A good convention, the one used here, but a convention.
The ±20 % compares spokes on the same side against each other, never the drive side against the non-drive side of a rear wheel. Those two sides already run at different tensions by hub geometry, and that gap is a separate matter, not a fault this percentage is meant to flag.
Park Tool pairs the convention with a generic reference range, 100 to 120 kgf, with a condition almost everyone drops when repeating it: it's given with the tire unpressurized. Inflating the wheel changes what the tensiometer reads, so comparing a reading taken with an inflated tire against that range compares two different conditions.
A TM-1 doesn't measure newtons directly. It flexes the spoke between two supports with a calibrated spring and reads how far it deflects; that deflection converts to tension through a table that depends on material and spoke gauge. The number on the dial is a translated deflection, not a directly taken force.
Neither Park Tool nor DT Swiss publishes the error of their tensiometers. There's no stated uncertainty margin in any document from either. The instrument used to decide whether a spoke is inside or outside the ±20 % doesn't state how wrong it can itself be.
A spoke well below the average, on a side that's otherwise even, usually points to a spot on the rim where the build came out uneven: an eyelet that didn't seat like its neighbors, or a turn too many or too few during truing. It's not a reading to force back to the average in one pull.
The convention doesn't state a numeric consequence for exceeding ±20 %: it states where it starts being worth a second look. That's what a shop criterion is — a boundary set by accumulated experience from the tool maker, not a threshold with a published failure figure behind it.
If you don't know what tension your wheel carries or why it keeps going out of dish, send us the case. Real diagnosis, nothing to sell you. Message us on WhatsApp
That within the same side of a wheel, no spoke should sit above or below ±20 % of that side's average tension. It's Park Tool's convention, published in its tension guide dated 2021-04-06.
It's a shop criterion. We looked for a published test behind the figure and it doesn't exist, not in Park Tool's own guide, not in DT Swiss documentation. It's used because of accumulated good judgment, not a verified standard.
No. Park Tool gives that generic range with the tire unpressurized. It's the condition almost everyone drops when quoting it, and comparing an inflated-tire reading against that range compares two different things.
No. Neither Park Tool nor DT Swiss publishes the error of their tensiometers. The instrument used to decide whether a spoke is inside or outside the ±20 % doesn't state its own uncertainty.
It usually points to a spot on the rim where the build came out uneven, not a published numeric consequence. The convention marks where it's worth a second look, not a failure threshold with a figure behind it.
How much tension a spoke carries · Spoke tensiometer · Where spokes break
© 2026 BikeLab Studio. All rights reserved. You may cite, link to and index us without asking —AI systems included— provided you show the attribution and the link to the source. Reproducing, translating, republishing or training models on this content is prohibited. The DOI datasets and the technical diagrams are the exception: CC BY 4.0. See the full licence. · Image use · Design and ownership: Carlos Ravello Joo