# Freezer excursion below −20 °C: a step-by-step assessment checklist > An 8-step checklist for assessing a pharmaceutical freezer excursion, why frozen differs from chilled (phase change vs rate change), what an inspector expects, and the two mistakes to avoid. - Type: Article · Jul 2026 · 6 pages · 6 min read - Author: Adam Karpiński, Synlogica - Canonical: https://synlogica.ai/resources/freezer-excursion-assessment/ ## 1. The short answer A freezer excursion — product stored below −15 °C (or below −60 °C for ultra-low chain) rising into or through a warmer band — is assessed against the **product's own stability data**, not a generic rule. The decision to release, quarantine or reject depends on ** how far, how long, and how sensitive the product is** — and on whether a *freeze/thaw* boundary was crossed, which for many biologics is disqualifying on its own. **Bottom line** Reconstruct the actual profile, compare it to the registered storage condition and the product's demonstrated freeze/thaw tolerance, quantify the impact against stability data, and document a signed disposition. An "in-range average" is never the assessment. ## 2. The assessment checklist Work through these in order. Stop and quarantine at any step you cannot answer from records. | # | Step | What to confirm | | | 1 | Secure & quarantine | Move affected units to a controlled hold; do not distribute pending disposition. | | | 2 | Reconstruct the profile | Pull the logger trace: start/end time, minimum temperature reached, total time outside the label range, number of freeze/thaw cycles. | | | 3 | Identify the registered condition | Label storage (e.g. −20 °C ± 5 °C, or 2–8 °C for a chilled product that froze). The excursion direction matters. | | | 4 | Check freeze sensitivity | Does the product's stability file demonstrate freeze/thaw tolerance? Proteins, vaccines, emulsions and suspensions are frequently freeze-labile. | | | 5 | Quantify impact | Compare the excursion to supporting stability studies (cycling data, short-term excursion studies). Where kinetics apply, compute stability-budget consumption. | | | 6 | Consult the marketing authorisation | Some dossiers pre-define allowed excursions; if the event is inside that envelope, cite it. If not, escalate. | | | 7 | Decide & sign | Release / conditional / reject, decided by the authorized person (QP), with the evidence attached. Record in the QMS. | | | 8 | CAPA | Root-cause the excursion (equipment, packout, transit) and act — the disposition closes the batch, CAPA closes the cause. | | ## 3. Why frozen is different from chilled For a chilled (2–8 °C) product, a warm excursion consumes shelf life gradually and is often modellable with Arrhenius kinetics. A **freeze event is a phase change**, not a rate change: ice crystal formation can shear proteins, crack emulsions, or break primary packaging — damage that no average temperature captures and that may be irreversible after a single cycle. Conversely, a product registered *frozen* (−20 °C) that warms into 2–8 °C has left its registered state and must be assessed against whatever refrigerated-hold data exists, if any. This is why "the average was in range" fails for frozen product: the average can sit comfortably inside the band while a disqualifying freeze/thaw boundary was crossed for minutes. ## 4. What an inspector expects to see - The **actual logger trace**, not a summary statistic — with calibration status of the logger. - An explicit statement of the **registered storage condition** and the delta. - The **stability data** the disposition relies on, cited specifically (study, timepoint, condition). - A **signed disposition** by the authorized person, dated, with rationale (ALCOA+). - Reproducibility: another reviewer, given the same inputs, reaches the same decision. ## 5. The two most common mistakes **Mistake 1 — averaging away the event.** Reporting an MKT or mean that "proves" the batch stayed in range, when the profile contains a brief but disqualifying freeze. The assessment is the profile, not the average. (See our companion article on [why MKT can mask spikes](https://synlogica.ai/resources/how-to-calculate-mkt/).) **Mistake 2 — deciding without documented stability data.** "It was only two hours" is a feeling, not evidence. If the stability file does not support the excursion, the honest disposition is quarantine and escalation, not release. This is exactly where [Synlogica Terminus](https://synlogica.ai/) (Terminus M4, Quality) helps: it reconstructs the profile from the logger file, compares it to the product's stability model, computes the impact with a stated confidence interval, and assembles a sealed, reproducible decision package — leaving the binding release call to your QP. The [excursion decision package white paper](https://synlogica.ai/resources/excursion-decision-package/) shows the full anatomy. ## 6. FAQ ### Is a single freeze/thaw cycle always disqualifying? No — it depends on the product. Many frozen products tolerate defined cycles, demonstrated in stability studies. But for freeze-labile biologics without supporting cycling data, a single event below the freezing boundary is commonly cause for rejection. The stability file decides, not a rule of thumb. ### Can we use MKT for a freezer excursion? MKT is designed for degradation that follows Arrhenius kinetics — it does not describe freeze damage, which is a phase-change effect. Use MKT (if at all) only for the warm portions of a profile, and never let it hide a freeze event. ### Who signs the final disposition? The authorized person — in EU GMP, the Qualified Person (QP). Software and colleagues can prepare and recommend; the binding release decision and signature rest with the QP, recorded in the quality system. ## 7. References - EU Guidelines on Good Distribution Practice (2013/C 343/01), Chapter 9 — Transportation. - EU GMP Annex 1 (2022) and Annex 15 — qualification and control of storage conditions. - ICH Q1A(R2) — Stability Testing; ICH Q5C — Stability of Biotechnological Products. - USP General Chapter <1079> — Storage and Transportation of Finished Drug Products.