Contents
1. The short answer
Mean Kinetic Temperature (MKT) is the single steady temperature that would cause the same amount of thermal degradation as the varying temperatures your product actually experienced. It is computed from the Arrhenius relationship, so warm periods count more than cool ones — which is why MKT is always equal to or higher than the arithmetic mean of the same readings.
2. The formula
The USP <1079> / Haynes formulation:
ΔH = heat of activation — 83.144 kJ/mol by convention (USP <1079>), unless the product's own value is known
R = 8.314 J/(mol·K) · Tᵢ = each reading in kelvin (°C + 273.15) · n = number of readings
Three practical rules: readings must be equally spaced in time (or the terms must be time-weighted), temperatures must be in kelvin, and one ΔH must be used consistently across the whole series.
3. Worked example — 24 h with a warm window
A refrigerated shipment (2–8 °C label) logs eight readings at 3-hour intervals. Mid-route, the trailer warms briefly:
| Reading | °C | T (K) | e−ΔH/RT |
|---|---|---|---|
| 1 | 5 | 278.15 | 2.435 × 10⁻¹⁶ |
| 2 | 5 | 278.15 | 2.435 × 10⁻¹⁶ |
| 3 | 6 | 279.15 | 2.769 × 10⁻¹⁶ |
| 4 | 7 | 280.15 | 3.147 × 10⁻¹⁶ |
| 5 | 12 | 285.15 | 5.885 × 10⁻¹⁶ |
| 6 | 9 | 282.15 | 4.053 × 10⁻¹⁶ |
| 7 | 6 | 279.15 | 2.769 × 10⁻¹⁶ |
| 8 | 5 | 278.15 | 2.435 × 10⁻¹⁶ |
Average of the eight exponentials: 3.241 × 10⁻¹⁶. Feed it back through the formula:
Compare: the arithmetic mean of the same readings is 6.88 °C. The single 12 °C reading contributes almost 2.4× the degradation weight of a 5 °C reading — that is the Arrhenius weighting doing its job. On longer profiles with sharper spikes, the gap between mean and MKT widens dramatically.
4. The mistake almost everyone makes
Averaging the temperatures and calling it MKT. The exponential must be applied per reading, before averaging. A spreadsheet that computes AVERAGE(°C) and labels the column "MKT" will understate thermal stress on every profile that contains an excursion — precisely the profiles where the number matters.
Second place: mixing units — ΔH in kJ/mol against R in J/(mol·K) shifts the result by three orders of magnitude, which is usually caught, or leaving temperatures in °C, which is not always caught and quietly produces nonsense.
5. When MKT is the wrong tool
MKT compresses a whole profile into one number, so it can mask short, sharp spikes — and regulators know it. EU GDP guidance and inspectors expect excursion assessment to consider the actual profile, not an MKT that "averages away" a freeze event or a brief 25 °C peak. MKT also assumes degradation follows Arrhenius kinetics with a single activation energy — false for freeze-sensitive biologics, where a single excursion below 0 °C can be disqualifying regardless of any average.
For a release decision after an excursion, the operational question is not "what was the MKT?" but "how much of this product's stability budget did the excursion consume?" — the product's own kinetics, integrated over the actual sensor trace, with stated uncertainty. That is the calculation Synlogica Terminus performs in its Quality module (Terminus M4), returning a sealed, audit-ready decision package in under a minute. The mathematics behind it is in our Arrhenius shelf-life modelling white paper.
6. FAQ
Which ΔH (activation energy) should I use?
USP <1079> uses 83.144 kJ/mol as the standard convention for storage and transit assessment. If the product's own activation energy has been characterised in stability studies, the product-specific value is more defensible — document which one you used and why.
Can MKT justify a temperature excursion on its own?
Generally no. An in-range MKT is supporting evidence, not a verdict: inspectors expect the actual excursion (duration, peak, product sensitivity) to be assessed against the product's stability data. MKT that hides a spike or a freeze event will not survive scrutiny.
Why is MKT always at or above the arithmetic mean?
Because the Arrhenius exponential grows faster than linearly with temperature, warm readings pull the weighted average up more than cool readings pull it down. The two are equal only when every reading is identical.
7. References
- USP General Chapter <1079> — Risks and Mitigation Strategies for the Storage and Transportation of Finished Drug Products.
- Haynes J.D., Worldwide virtual temperatures for product stability testing, J. Pharm. Sci. 60 (1971).
- ICH Q1A(R2) — Stability Testing of New Drug Substances and Products.
- EU Guidelines on Good Distribution Practice of Medicinal Products for Human Use (2013/C 343/01), Chapter 9.