Knowledge of the stability of serum samples stored in large biobanks is pivotal for reliable assessment of hormone-dependent disease risks. We studied the effects of sample storage time and season of serum sampling on the stability of 25-hydroxy vitamin D (25-OHD) and androstenedione in a stratified random sample of 402 women, using paired sera from the Finnish Maternity Cohort. Serum samples selected were donated between 6 and 24 yr ago. The storage time did not affect serum 25-OHD and androstenedione levels. However, there was a significant mean difference in the 25-OHD levels of sera withdrawn during winter (first sample) vs. during summer (second sample; -18.4 nmol/l, P <or= 0.001). Also at the individual level, there were significant differences in average 25-OHD levels between individuals with the paired sera taken at winter-winter compared with other alternatives (summer-winter, winter-summer, and summer-summer). The androstenedione levels showed no such differences. Long-term storage does not affect serum 25-OHD and androstenedione levels, but sampling season is an important determinant of 25-OHD levels. Stored serum samples can be used to study disease associations with both hormones. However, sampling season needs to be taken into account for 25-OHD by considering matching and stratification and, if possible, serial sampling.