The ability of an albumin-to-creatinine ratio, measured in a single untimed urine specimen, to indicate the likelihood of developing overt diabetic nephropathy was determined in 439 Pima Indians (134 men, 305 women) aged 25 years or older with non-insulin-dependent diabetes. During a mean follow-up period of 4.2 years, 59 (13%) of the subjects developed overt nephropathy, 47 (80%) of whom had albumin-to-creatinine ratios of 30 mg/g or greater at baseline. Subjects with albumin-to-creatinine ratios of 30 to 299 mg/g (a level of excretion often termed "microalbuminuria") had 9.2 times (95% confidence interval, 4.4 to 21.4) the incidence of overt nephropathy of those with ratios of less than 30 mg/g. Furthermore, the albumin-to-creatinine ratio remained a strong predictor of overt nephropathy even when controlled for age, sex, diabetes duration, mean blood pressure, and 2-hour postload plasma glucose concentration with a proportional-hazards function analysis. Thus, an albumin-to-creatinine ratio measured in a single untimed urine specimen is an effective means of identifying diabetic subjects who are at risk of developing overt nephropathy that could replace the more traditional timed urine collections.