Retrospective studies of patients with non-insulin-dependent diabetes mellitus (NIDDM) have suggested that microalbuminuria predicts early all-cause (mainly cardiovascular) mortality independently of arterial blood pressure. These findings have not been confirmed in prospective studies, and it is not known whether the predictive power of microalbuminuria is independent of other major cardiovascular risk factors. During 1985-1987, we examined a representative group of 141 nonproteinuric patients with NIDDM for the prevalence of coronary heart disease and several of its established and putative risk factors, including raised urinary albumin excretion (UAE) rate. Thirty-six patients had microalbuminuria (UAE 20-200 micrograms/min), and 105 had normal UAE (less than 20 micrograms/min). At follow-up, an average of 3.4 yr later, 14 patients had died. There was a highly significant excess mortality (chiefly from cardiovascular disease) among those with microalbuminuria (28%) compared to those without microalbuminuria (4%, P less than 0.001). In univariate survival analysis, significant predictors of all-cause mortality included microalbuminuria (P less than 0.001), hypercholesterolemia (P less than 0.01), hypertriglyceridemia (P less than 0.05), and preexisting coronary heart disease (P less than 0.05). The predictive power of microalbuminuria persisted after adjustment for the effects of other major risk factors (P less than 0.05). We conclude that microalbuminuria is a significant risk marker for mortality in NIDDM, independent of the other risk factors examined. Its presence can be regarded as an index of increased cardiovascular vulnerability and a signal for vigorous efforts at correction of known risk factors.