Paddle Surface Roughness: The Rz Limit and How It's Tested

A USA Pickleball approved paddle face averages no more than 30 micrometres Rz and 40 micrometres Rt, returns a kinetic coefficient of friction of 0.1875 or lower, and reflects no more than 80 gloss units at 60 degrees. None of those four numbers appears in the Rulebook. The 2026 Official Rulebook bans rough texturing, sandpaper characteristics and anti-skid paint in words, and Rule 3.D.4 then sends the actual criteria to the Equipment Standards Manual, where clause 2.E.2.a.1 turns the words into readings.

How the reading is taken decides as much as the limit does. Each face is measured six times, in six different directions, and averaged on its own. No single one of those readings may pass 33 micrometres Rz or 44 micrometres Rt. A face can average 28 micrometres and still fail on one spike.

First, a third-party primer on the parameters.

The Four Numbers a Legal Paddle Face Has to Clear

Rule 3.D.5 of the 2026 Rulebook is a list of words. It prohibits holes, cracks, delamination, rough texturing, sandpaper characteristics, rubber, anti-skid or sand-textured paint, anything else that lets a player impart excessive spin, and a reflective finish that might affect an opponent's vision. Not one line carries a number. Rule 3.D.4 explains why: paddle material criteria are the ones "specified in the USA Pickleball Equipment Standards Manual".

The manual, Revision 3.0 of January 2025, calls itself the flow down from the Rulebook to "specific definition of the testing that is done to confirm compliance". Clause 2.E.2 repeats the prohibition, then adds the part that makes it enforceable. The surface is judged by roughness, coefficient of friction and coefficient of restitution measurements. Three tests, three sets of numbers, and a gloss ceiling in 2.E.2.b on top. The third of those tests, coefficient of restitution, is the power limit covered in our explainer on the PBCoR standard. In millimetres the caps read 0.030 mm Rz and 0.040 mm Rt, a scale a machinist recognises and a paint shop rarely does.

Read the table as four independent gates. A face that clears roughness can still fail friction: different protocols, different questions.

Source: USA Pickleball Equipment Standards Manual, Revision 3.0 (January 2025), clauses 2.E.2.a.1, 2.E.2.a.2 and 2.E.2.b, read 12 September 2026.

Surface property	Average limit per face (µm, ratio or GU)	Single-reading cap (µm)	Clause	Verdict: what fails it
Roughness Rz, peak to valley	30 µm	33 µm	2.E.2.a.1	A high average, or one spike above the cap
Roughness Rt, peak to valley	40 µm	44 µm	2.E.2.a.1	Same six readings, second parameter, judged the same way
Kinetic coefficient of friction	0.1875	none stated	2.E.2.a.2	Any value above it, whatever the finishing process
Gloss measured at 60 degrees	80 GU	none stated	2.E.2.b	A glossy edge guard, not only a glossy face

Rz, Rt, Ra, Sa: Which Parameter the Standard Actually Caps

Roughness is not one number. It is a family of parameters, and the manual caps two. Clause 2.E.2.a.1 names Rz, glossed there as "average maximum height, peak to valley", and Rt, "maximum height, peak to valley". Both describe how far the highest peak sits above the deepest valley. Neither describes average deviation, which is what Ra reports.

That difference is why a lab report can be accurate and useless at once. Arithmetic mean height and total height sit as separate parameters in clause 4.2 of ISO 21920-2:2021, the standard that defines profile surface texture today. A smooth-looking face with a few tall, widely spaced peaks can hold a low Ra and still break the single-reading ceiling of 0.033 mm. Peaks are what the sport regulates. Peaks are what grab the ball.

Two version facts belong on the spec sheet. ISO 21920-2:2021 cancels and replaces ISO 4287:1997, and Germany's metrology institute records the ISO 21920 series as replacing ISO 4287 and ISO 4288. The new edition relates all field parameters to the evaluation length. Two reports quoting the same parameter under different editions were not computed the same way. Name the edition, or the number is half a fact.

Source: parameter names and families from ISO 21920-2:2021 clause 4.2 and ISO 25178-2:2021 as cross-referenced there; capped values from Equipment Standards Manual 2.E.2.a.1. Read 12 September 2026.

Parameter	What it reports	Defined in	Verdict for a submission report
Rz	Peak-to-valley height, profile method	ISO 21920-2:2021	Required. This is the capped average of 30 µm
Rt	Total peak-to-valley height, profile method	ISO 21920-2:2021	Required. This is the capped average of 40 µm
Ra	Arithmetic mean height, profile method	ISO 21920-2:2021	Not accepted. The manual caps peaks, not averages
Sa	Arithmetic mean height of an areal scan	ISO 25178-2:2021	Not accepted. Different standard, different quantity

Averaged limit versus single-reading ceiling, per paddle face. Source and method: Both series are T1 primary-document values transcribed verbatim from the USA Pickleball Equipment Standards Manual, Revision 3.0 (January 2025), clause 2.E.2.a.1 (grounding ids esm-rz-rt-limits and esm-single-reading). No value is measured in-house, modelled or interpolated.

Capped roughness parameter	Average limit per face (µm)	Single-reading ceiling (µm)
Rz	30	33
Rt	40	44

Six Directions, Both Faces, and the Single Reading That Fails a Face

Clause 2.E.2.a.1 sets the method as tightly as the limit. A Starrett SR160 or SR300 surface roughness tester, or an equivalent, takes readings "in six different directions". The six values are averaged for that face, and each face has to meet the requirement on its own average. The data set is those six points, and no one of them may exceed 33 micrometres Rz or 44 micrometres Rt, which is 0.033 mm and 0.044 mm.

Three consequences follow, and a production spec usually misses them. Direction matters: a woven or peel-ply pattern can read differently across the weave than along. The face is the unit, so a paddle with a printed logo on one side and bare weave on the other is two measurements. And the spike rule beats the average, so one embedded particle can fail a face that reads low five times out of six.

Friction is a second, unrelated test. Clause 2.E.2.a.2 runs it in accordance with ASTM D1894-14, a method written to determine the coefficients of starting and sliding friction "of plastic film and sheeting", and passes a paddle at a kinetic coefficient of 0.1875 or lower. Two things follow from that scope. The number is a comparison against a fixed protocol, not a property of the paddle. The edition is specific too: the active edition of D1894 is the 2024 revision, while the manual still names the 2014 one. Testing runs through labs hosted at Element U.S. Space and Defense, reviewed by the Equipment Evaluation Committee.

Ask for the six raw readings per face, not the averages. An average hides the spike that fails a lot.

Where the Reading Moves Between Your Golden Sample and the Container

A passed sample is a snapshot of one face on one day. Clause 2.F.2 subjects manufacturers to ongoing compliance testing so that "all products as produced and sold are identical to the paddle that has been tested/certified". Clause 2.F.2.a goes further. Where a design has break-in periods or gradual change, the paddle must stay compliant across its entire useful life, "inclusive of any degradation to benefit". A finish that gets grippier as the resin wears is a compliance problem, not a feature.

Changing the finish is also a paperwork event. Clause 2.E.7 requires a unique name or number for paddles with a different core material, surface material or other significant differences. Adding a coating to an approved raw face creates a new model, not a new batch. Clause 2.E.6 closes the other door. Anti-skid paint, paint textured with sand, rubber or vinyl compounds, rubber surfaces and sandpaper characteristics are prohibited outright, and each is judged by the same three measurements.

Approval can also be withdrawn. Clause 2.F.1 allows revocation on 18 months' notice where equipment was materially changed or degrades under ordinary use. Specifications move through a Notice of Proposed Rulemaking to manufacturers, so a limit locked two seasons ago deserves a re-read before the next run.

This is where a texture decision stops being a styling choice. We build raw T700 carbon and fiberglass faces on 13 mm and 16 mm polypropylene honeycomb cores, with custom surface texture, through the OEM paddle programme, and the finishing step is the variable that moves a reading. Which texture to choose is a separate argument, made in our guide to paddle grit and spin. This page is about whether the face you chose can be proved legal.

The Spec-Sheet Lines, and What We Check Before a Face Ships

Most surface specifications say "USAPA compliant texture" and stop. That line cannot be checked, and a report written against it settles nothing. Seven lines make the reading verifiable, and every one comes out of clause 2.E.2. They slot into the surface block of a paddle specification sheet.

• Parameters: Rz and Rt, reported together. Ra or Sa alone is not a result against this specification.
• Limits: average 30 µm Rz and 40 µm Rt per face, no single reading above 33 µm Rz or 44 µm Rt.
• Sampling: six readings in six different directions on each face, averaged per face, both faces reported separately.
• Instrument: Starrett SR160 or SR300, or an equivalent named in the report.
• Friction: kinetic coefficient of friction of 0.1875 maximum, tested in accordance with ASTM D1894-14 as the manual names it.
• Gloss: 80 GU maximum at 60 degrees, on the edge guard as well as the face.
• Editions: name the manual revision and date, and the ISO edition the parameters were computed under.

What we document, and what we do not publish

Our side of that block is what a buyer can verify before a run starts. Full spec documentation is available, and every paddle passes QC before it leaves the line. The finish is fixed at golden-sample approval, not agreed in a chat thread, and the minimum is 50 paddles.

What we do not publish is an in-house Rz reading. We have not measured one under the six-direction protocol, and an invented number is worth less than none. Read every figure here, from 0.030 mm upward, as the published limit it is.

Conclusion

Write the surface line as numbers and a method, not as a claim of compliance: 30 and 40 micrometres averaged per face, 33 and 44 as single-reading ceilings, 0.1875 kinetic friction, 80 GU of gloss, six directions on each face. The answer flips only when USA Pickleball moves the specification through a Notice of Proposed Rulemaking, so the revision and its date belong in the same sentence as the limit.

Before you ask a factory to price a textured face, have three things ready: the parameter pair you will accept in a report, the edition of each standard you are citing, and whether both faces are to be reported separately. Those three answers turn a texture conversation into a specification you can both check against the same document. The core-side equivalent sits in our rulebook decoder for paddle core legality.

Frequently Asked Questions