The effects of storage for 6 months or 2 years at -20 degrees C were studied in urine samples from Type II diabetic patients by assaying albumin by immunoturbidity, N-acetylglucosaminidase (EC 3.2.1.30) by methoxynitrovinylphenol release, and creatinine by the Jaffé method. There were significant decreases (P < 0.001) in albumin/creatinine ratios from 1.14 (0.63-2.98) to 0.83 (0.32-2.12) g/mol (median + interquartile ranges) after 6 months (n = 97), and from 1.64 (0.74-5.72) to 1.00 (0.37-4.54) g/mol after 2 years (n = 89). The percentage of samples with results below the detection limit of the albumin assay (2 mg/L) increased from 5% to 21% after 6 months and from 0% to 34% after 2 years. N-Acetylglucosaminidase/creatinine ratios decreased (P < 0.001) from 520 (358-832) to 380 (263-695) U/mol after 6 months and from 520 (330-865) to 258 (82-462) U/mol after 2 years. The effect of storage was greater in samples with concentrations in the normal range (< 2.5 g/mol for albumin/creatinine, < 500 U/mol for N-acetylglucosaminidase/creatinine). Samples with albumin concentrations more than twice the normal range were still detected as abnormal after storage at -20 degrees C; e.g., 18% were > 5 g/mol (albumin/creatinine) initially, with 17% > 5 g/mol after 6 months of storage. We therefore recommend storage of urine samples at 4 degrees C for no longer than 7 days before assay.