In most survival studies in NIDDM, microalbuminuria (urinary albumin excretion rate 20-200 microg/min) predicts early mortality; in cross-sectional studies, it is associated with coronary heart disease (CHD) morbidity. It is unclear, however, whether microalbuminuria is a risk factor for the development of CHD or the result of it, and little is known of the factors that predispose to the development of microalbuminuria in NIDDM. We examined these issues in a 7-year prospective study of a hospital-based cohort comprising 146 white NIDDM patients without clinical albuminuria. Microalbuminuria was a significant risk factor for both all-cause mortality (relative risk 3.94, 95% CI 2.04-7.62) and CHD mortality (relative risk 7.40, 95% CI 2.94-18.7) when adjusted for age only. Its independent predictive power did not persist, however, in age-adjusted multivariable survival analysis that allowed for the other significant risk factors: male sex, preexisting CHD, high levels of glycated hemoglobin, and high serum cholesterol. Among men free of CHD at baseline, the independent risk factors for CHD morbidity and mortality were microalbuminuria, current smoking, high diastolic blood pressure, and high serum cholesterol (all P < 0.05). For the 100 NIDDM patients with normoalbuminuria at baseline, the incidence of microalbuminuria was 29% over the 7-year period. In that group, fasting plasma glucose, current smoking, preexisting CHD, and high initial urinary albumin excretion rate were risk factors for the development of microalbuminuria (all P < 0.05). When men and women were analyzed separately, preexisting CHD was a significant risk factor in men only. These results demonstrate that microalbuminuria predicts incident clinical CHD in men with NIDDM. Preexisting CHD is also a risk factor for incident microalbuminuria in men, however, suggesting that microalbuminuria and CHD are not causally related but rather reflect common determinants.