Epidemiological studies on folate and chronic diseases often involve the use of frozen serum stored in biorepositories for decades. Folate instability may attenuate associations between folate status and study endpoints. In this cross-sectional study, we retrieved serum samples stored at -25 degrees C for 0, 4, 6, 17, or 29 y in the Janus biobank. Samples were obtained from a total of 650 men aged 40-49 y at the time of blood collection and were evenly distributed according to storage time. Folate was determined by a liquid chromatography tandem MS (LC-MS/MS) assay that measures 5-methyltetrahydrofolate (5mTHF), its oxidation product 4-alpha-hydroxy-5-methyltetrahydrofolate (hmTHF), and other folate species; by a Lactobacillus rhamnosus microbiological assay; and by LC-MS/MS as p-aminobenzoylglutamate (pABG) equivalents after oxidation and mild acid hydrolysis of the folate species. Concentrations of 5mTHF and microbiologically active folate were lower in samples that had been subjected to long-term storage and the data were consistent with a decrease of 3.2 and 2.8%/y, respectively. hmTHF was detected in all specimens but did not accumulate upon long-term storage (>4 y). Folate measured as pABG declined at a slow rate of 0.98%/y and approximately 80% of the folate was recovered after 29 y of storage. B-vitamin status did not differ between individuals delivering samples at different time points, as assessed by measuring total homocysteine, methylmalonic acid, and serum vitamin B-12. In conclusion, folate is substantially degraded in serum frozen for decades but can to a large extent be recovered as pABG equivalents. The pABG assay appears to be the method of choice for the determination of folate status in archival serum samples.