Calibration Programs: Why Your Thermometers Might Be Lying to You

By GIFSQ | Published October 21, 2026

A Two-Degree Error That Can Make People Sick

Picture this: your cook-chill line shows product hitting 74°C on the display — safely above your 70°C critical limit. Except the probe reads two degrees high, and the actual product temperature is 72°C. Your CCP logs are perfect, your product is compliant on paper, and your real safety margin just evaporated. Nobody knows, because the only thing that could have told them is sitting in a drawer with an expired calibration sticker.

This isn’t hypothetical. Calibration failures show up in audit nonconformities with depressing regularity, and the reason is simple: every food safety decision you make — cooking, cooling, freezing, holding — starts with a measurement. A wrong measurement poisons every decision built on it. Your HACCP critical limits, your FSMA preventive controls, your cold-chain records are only as trustworthy as the instruments behind them.

ISO 22000:2018 clause 7.1.5 requires measuring equipment to be calibrated or verified at specified intervals against measurement standards traceable to international or national standards. 21 CFR 117.40(a)(1) requires instruments used for monitoring be “accurate and precise and adequately maintained.” Calibration isn’t a nice-to-have quality activity — it’s a legal requirement.

What Belongs in a Calibration Program

A real calibration program has five components, and a sticker on the instrument is only one of them:

• An instrument register: every device that produces a measurement your food safety system relies on — dial and digital thermometers, temperature data loggers, pH meters, scales, metal detector test pieces, pressure gauges, aw meters, chlorine test kits, refractometers. If a decision depends on it, it’s in the register.
• Calibrated ranges and tolerances: define the operating range and acceptable tolerance for each instrument. A thermometer accurate to ±2°C at 100°C may not be accurate to ±2°C at 0°C. Your tolerance should connect to your critical limits — if your CCP limit is 70°C and your thermometer is only accurate to ±2°C, you need to account for that.
• Intervals: risk-based frequency. Thermometers used at CCPs typically get checked daily or before each use (ice-point check) plus full calibration on a scheduled interval (quarterly to annually). ISO 17025-accredited labs provide traceable calibration certificates; interim checks are done in-house.
• Traceability: your calibration must trace back to national or international standards. That chain — NIST in the US, NRC in Canada, PTB in Germany, through an ISO 17025-accredited lab to your instrument — is what makes the certificate mean something.
• Records: as-found and as-left readings, date, who did it, the reference standard used, and the next due date. Records of the 5 years typically needed for a full food safety audit cycle.

Interim in-house checks are the program’s early warning system. An ice-point check takes 60 seconds: crushed ice, distilled water, stir, wait for stabilization, read. Pure water ice point is 0°C (32°F). If your probe reads 1.5°C, you have a 1.5°C bias — log it, correct for it, and flag the instrument for full calibration if it’s outside your tolerance.

The Ice-Point Check: 60 Seconds That Catch Drift

Here’s the procedure, exactly as it should appear in your work instruction:

1. Fill an insulated cup with crushed ice, then add distilled water until the cup is full.
2. Stir the mixture for 30 seconds to create a uniform slurry.
3. Immerse the probe at least 5 cm (2 inches) into the center of the slurry — not touching the sides or bottom.
4. Wait 30 seconds for the reading to stabilize, then record it.
5. Compare to 0°C. If outside your defined tolerance (commonly ±0.5°C for food probes), remove the instrument from service, tag it, and route it for calibration or replacement.
6. Log the result with date, instrument ID, and operator initials.

Do the same at the boiling point (100°C at sea level, corrected for altitude — roughly 1°C lower per 300 m) for a two-point verification. Boiling-point checks catch span errors that ice-point checks miss.

What Happens When Calibration Fails

This is the part most programs get wrong. When an instrument is found out of tolerance, you don’t just recalibrate it — you have to assess the product impact. ISO 22000 requires you to evaluate the validity of previous measurement results. In practice, that means:

Determine the last known-good date — when was this instrument last verified in tolerance? Every measurement between that date and the failure is suspect. For a CCP thermometer, that means every batch whose compliance was verified by that instrument needs review. Depending on what you find, that review ranges from “the readings were well inside the limit, product is safe” to a hold, investigation, and possible recall.

This is why calibration intervals and interim checks matter commercially, not just technically. A thermometer that drifts out of tolerance between annual calibrations creates a year of suspect product. One that gets a 60-second ice-point check before each use creates a single shift of exposure. The cost of the interim check is essentially zero; the cost of the exposure can be the entire business.

Also remember: calibration covers more than thermometers. A scale that’s wrong affects your allergen dosing and your preservative concentrations — both safety-critical. Metal detector test pieces must be certified to the correct size and ferrous/non-ferrous/stainless composition. pH meters drift with electrode age and need buffer verification before use. Each one is a separate line in your register, with its own interval and its own tolerance.

Don’t overlook the “hidden” instruments either. The data logger in the blast chiller that nobody downloads, the infrared thermometer the receiving dock uses on every delivery, the timer on the smokehouse that the whole thermal process assumes is accurate — these are measurement devices your food safety system depends on, and they drift like everything else. Infrared thermometers deserve special caution: they’re surface-only, emissivity-dependent, and useless on shiny or reflective surfaces unless properly set up. Many plants use them for quick checks but make CCP decisions from them anyway. If an infrared gun is used for any decision that matters, it belongs in the register with its own tolerance and its own verification procedure — or it gets replaced by a probe thermometer for the critical measurements.

Monday morning: pull your calibration register. Find every instrument whose sticker date has passed. Now ask the uncomfortable question for each one: how much product has it measured since the last known-good check? The answer sets your priority list — and your argument for daily interim checks on the critical ones.