The 2026 USP <41> Revision: Five Changes Your Lab Can't Afford to Miss

For years, USP <41> was refreshingly simple. Put a test weight on the balance, confirm the reading falls within 0.10% of nominal, document it, move on. That was essentially the whole requirement.

The February 2026 revision changed that. It's not a cosmetic update — it adds an entirely new calibration section, tightens test weight selection, and introduces a requirement that will appear on every calibration certificate going forward. If your SOPs still reflect the 2019 chapter, you have a gap that an auditor will find.

Here's what actually changed, and what it means for your weighing program.

First, a quick orientation: who USP is and why this matters

USP is the rulebook writer. The FDA is the enforcer. USP itself doesn't police laboratories, but its standards get cited in warning letters and 483s — and USP standards are referenced globally, not just in the US.

One distinction worth internalizing: chapters numbered below 1000 are compendial standards containing enforceable requirements. Chapters above 1000 are guidance — best practices and recommendations.

That's why the 2026 revision matters so much. It pulled concepts that previously lived in the guidance chapter <1251> into the enforceable chapter <41>. What used to be "recommended" is now "required."

Change #1: Explicit calibration requirements were added

Before (2019): The chapter said little more than that the test weight must fall within the balance's operating range.

Now (2026): There's an entirely new CALIBRATION section with specific expectations:

• Calibration must establish metrological traceability to SI units
• Calibration results must include measurement uncertainty
• Calibration must occur periodically and before/after any operation that may significantly affect performance
• Performance checks must occur between calibrations
• Calibration frequencies must be risk-based and defined in your quality system

That last point deserves attention. "We calibrate annually because we've always calibrated annually" is no longer a defensible position. Your frequency needs a documented rationale tied to risk.

Change #2: Calibration uncertainty is now mandatory (the headline change)

This is the one that catches labs off guard.

Measurement uncertainty must now appear on the calibration certificate. Not as an internal calculation, not as a value available on request — on the certificate.

The calculation itself involves the uncertainty of the reference weights, a coverage factor, and a combined-then-expanded uncertainty result. It's not conceptually difficult, but it's a new artifact that has to be produced consistently, documented correctly, and placed where an auditor can see it.

If your current certificates don't carry an uncertainty statement, that's the first gap to close.

Change #3: Repeatability test weight selection got restrictive

Before (2019): Essentially any suitable test weight within the operating range. No lower limit specified.

Now (2026): The test weight must:

• Be ≤ 5% of balance capacity
• Be a single denomination (no stacking combinations)
• Not be less than 100 mg

The single-denomination rule trips people up. If you've been building your repeatability test weight from a combination of smaller weights, that practice no longer complies. And the 100 mg floor means some historical low-end repeatability testing is now out of bounds.

Change #4: Minimum Weight vs. Smallest Net Weight, finally clarified

The 2019 chapter implied the distinction. The 2026 revision makes it explicit — and the two are frequently confused, so it's worth being precise.

Minimum Weight is measured. It's calculated from repeatability testing:

Minimum Weight = 2000 × standard deviation of repeatability results

It can drift over time as balance performance changes.

There's a wrinkle in the math, though. If your repeatability is perfect, your standard deviation is 0.00000 — which would make your minimum weight zero. That's obviously not possible. So USP specifies that if the standard deviation obtained is less than 0.41d (where d is the scale interval), you substitute 0.41d.

In practical terms, the minimum weight of any balance is effectively 820 divisions (820 × resolution):

A worked example on a 5-place balance: if your standard deviation is 0.000015 g, your minimum weight is 2000 × 0.000015 = 0.03000 g (30 mg). If your standard deviation comes back at 0.000000 g, you substitute the 0.41d floor: 2000 × 0.0000041 = 0.00820 g (8.2 mg).

Smallest Net Weight is different. It's user-defined — it reflects the smallest quantity you actually weigh in routine operation. It cannot be less than the minimum weight, and USP <1251> suggests building in a 1.5x to 2x safety factor above minimum weight.

Why the safety factor? Because minimum weight drifts. A balance that determines at 16 mg at installation may drift to 24 mg by the as-found calibration. If your smallest net weight sits right at the minimum weight, you're out of compliance the moment performance shifts. Set it at 30 mg with a 16 mg minimum, and you've built in room for reality.

Change #5: Multi-interval and multi-range balances now have rules

The 2019 chapter offered no guidance here at all. The 2026 revision addresses both — and the distinction matters because the testing requirements differ.

Multi-Interval balances change resolution automatically as weight increases. → Repeatability testing must occur in the finest weighing range.

Multi-Range balances have two or more defined, selectable ranges. → Testing is required in each weighing range used routinely. Preload may be needed to access coarser ranges.

Get this backwards and you've either done unnecessary testing or left a range untested.

Two more worth noting

Accuracy testing absorbed <1251> elements. Pass/fail on a single accuracy check is no longer the whole story. Sensitivity, eccentricity, and linearity are now part of the picture within <41> itself.

Manufacturing balances are explicitly out of scope. The new chapter states plainly that it does not cover balances used for manufacturing — clarifying that <41> is a laboratory-focused chapter.

What to do about it

A short, honest checklist:

1. Update your SOPs to reflect the new calibration section, including risk-based frequency rationale.
2. Confirm your certificates carry an uncertainty statement. If they don't, that's your most visible gap.
3. Audit your repeatability test weights against the ≤5%, single-denomination, ≥100 mg rules.
4. Verify your smallest net weight has an appropriate safety factor above your determined minimum weight.
5. Classify your multi-interval vs. multi-range balances and confirm testing matches the right rule.
6. Document training on the revised procedures.

Where Quantus fits

At Quantus, our balance calibration methodology already incorporated <1251>-level testing — sensitivity, linearity, and eccentricity — before the 2026 revision required it. For our customers, the transition wasn't a scramble. It was a documentation update.

We've built the uncertainty calculation directly into our validated testing tool, so the value that has to appear on your certificate is generated consistently every time, not reconstructed by hand. Our SOPs and forms are updated and released. Our service templates reflect the new requirements.

If you're not certain your weighing program meets the 2026 revision, we'd rather find out with you than have an auditor find out for you.

Schedule a calibration program review.