This study was undertaken to assess the usefulness of different techniques for determination of albumin excretion rate (AER). Ninety patients with type I (insulin-dependent) diabetes mellitus and 45 with type II (non-insulin-dependent) diabetes mellitus, with AER/24 h of less than 200 micrograms/min, were included. All patients were free of major systemic complications of diabetes and overt kidney disease (mean serum creatinine 1.1 +/- 0.1 mg/dl, range 0.4-1.2 mg/dl). We compared timed day, night, and 24-h specimens, as well as timed spot specimens during water-induced diuresis. Most patients with type I (75 of 90) and type II (30 of 45) diabetes had AER less than 20 micrograms/min and showed significant differences in AER that were dependent on the collection time. Differences were diminished or absent with AER less than 20 micrograms/min. Sensitivity, specificity, and prediction rates of AER in different specimens were evaluated against 24-h AER. Use of albumin concentrations and albumin-creatinine ratios did not improve test performance in comparison with AER. Sampling time and the overall rate of AER influenced measurement of urinary albumin excretion. Day or 24-h AER is most useful to determine the presence of abnormal AER. AER and albumin concentration in spot samples are of limited use for initial screening and frequently require day or 24-h specimens of AER for confirmation. Day or 24-h AER should be used for long-term follow-up of the diabetic patient.