Twenty-four-hour, four-hour (8 to 12 am), and overnight urine collections were examined for their ability to detect microalbuminuria in 292 patients with insulin-dependent diabetes mellitus (IDDM). Albumin excretion rate (AER) was measured and also estimated from the product of the urinary albumin/creatinine ratio (A/C) and the calculated 24-hour creatinine excretion. The fractional excretion of albumin (FEA) was also determined in aliquots from each urine sample. The correlation coefficients between measured 24-hour AER and estimated AER were 0.940 and 0.956 for four-hour and overnight collections, respectively (significance of each correlation, P less than 0.001). There was no advantage in using the FEA over the A/C ratio in predicting measured AER. Urinary A/C ratios (mg/mg) between 0.03 and 0.31 in the four-hour collections were highly predictive of microalbuminuria and of measured AER in the 24-hour collections: AER24-h (microgram/min/1.73 m2) = 2.74 + 0.870 x A/C4-h (all log10 values). In a subgroup of 175 patients having all three collections validated, 34 (20%) had microalbuminuria defined as AER 20 to 200 micrograms/min/1.73 m2 in at least two of the three samples and 44 (25%) had overt nephropathy (greater than 200 micrograms/min/1.73 m2). The ability of the AER in one urine collection to predict microalbuminuria in at least one of the other two collections was assessed in these 175 patients. Compared with the overnight urine collection, the four-hour collection had greater sensitivity while affording similar specificity and positive predictive value. Based on these data, the A/C ratio from a morning urine sample following initial AM voiding would seem adequate for the detection and monitoring of microalbuminuria in patients with IDDM.