Objectives Retrospective studies frequently assume analytes long-term stability at ultra-low temperatures. However, these storage conditions, common among biobanks and research, may increase the preanalytical variability, adding a potential uncertainty to the measurements. This study is aimed to evaluate long-term storage stability of different analytes at Methods Twenty-one analytes commonly measured in clinical laboratories were quantified in 60 serum samples. Samples were immediately aliquoted and frozen at <-70 degrees C, and reanalyzed after 11 +/- 3.9 years of storage. A change in concentration after storage was considered relevant if the percent deviation from the baseline measurement was significant and higher than the analytical performance specifications. Results Preanalytical variability (CVP) due to storage, determined by the percentage deviation, showed a noticeable dispersion. Changes were relevant for alanine aminotransferase, creatinine, glucose, magnesium, potassium, sodium, total bilirubin and urate. No significant differences were found in aspartate aminotransferase, calcium, carcinoembryonic antigen, cholesterol, C-reactive protein, direct bilirubin, free thryroxine, gamma-glutamyltransferase, lactate dehydrogenase, prostate-specific antigen, triglycerides, thyrotropin, and urea. As nonnegligible, CVP must remain included in reference change value formula, which was modified to consider whether one or two samples were frozen. Conclusions After long-term storage at ultra-low temperatures, there was a significant variation in some analytes that should be considered. We propose that reference change value formula should include the CVP when analyzing samples stored in these conditions.