The excretion of small quantities of urinary albumin (microalbuminuria = urinary albumin excretion rate, UAER = 20-200 micrograms/min) may predict renal function in both insulin-dependent and noninsulin-dependent diabetes. We compared radioimmunoassay with the immunoturbidimetric method to detect early increases in urine albumin concentration. More problems have been encountered in deciding which method of collecting urine best differentiate between early onset diabetic nephropathy and normality. Random urine samples collected at clinics are convenient but show wide variations in concentration and the effects of exercise. Such variations may be overcome by using a rest period and correcting for urine creatinine concentration. We studied 21 IDDM patients (12 female, 9 male), aged 13-33 years old (mean 21) and 11 nondiabetics (6 female, 5 male), aged 15-30 years old (mean 23). All gave negative results on testing with Albustix at clinic visits. All subjects passed urine immediately after they got up in the morning. The results disclosed: (1) The correlation coefficient of albumin excretion (micrograms/ml) in the urine collected overnight with that collected over 24 hours was good (r = 0.89, p less than 0.001). (2) When the albumin excretion rate of the urine collected overnight was expressed as microgram albumin/mg creatinine, the correlation was also as good as the 24-hr urine albumin excretion (microgram albumin/mg creatinine) (r = 0.87, p less than 0.001). (3) The results of our study support the use of urine samples collected overnight, corrected for creatinine, to estimate microalbuminuria.(ABSTRACT TRUNCATED AT 250 WORDS)