The impact of microalbuminuria and macroalbuminuria on mortality was evaluated prospectively in 328 Caucasian patients with non-insulin-dependent diabetes mellitus (NIDDM) followed for 5 years. One hundred ninety-one (109 men and 82 women) patients with normoalbuminuria (albumin excretion rate [AER] <30 mg/24 h), 86 (50 men and 36 women) patients with microalbuminuria (AER 30-299 mg/24 h), and 51 (43 men and 8 women) patients with macroalbuminuria (AER ≥300 mg/24 h) <66 years old at entry were followed from 1987 until death or until 1 January 1993. Mean age at entry was 54 (SD 9) years. In January 1993, 8% of patients with normoalbuminuria, 20% of patients with microalbuminuria, and 35% of patients with macroalbuminuria had died (predominantly from cardiovascular disease) (P < 0.01 [normoalbuminuria versus micro- and macroalbuminuria] and P < 0.05 [microalbuminuria versus macroalbuminuria]). Cox multiple regression analysis revealed significant predictors of all-cause mortality to be preexisting coronary heart disease (relative risk [95% confidence interval]), 2.9 (1.6-5.1); log10AER (factor 10), 1.9 (1.4-2.6); HbA(1c) level (%), 1.2 (1.0-1.4); and age (years), 1.08 (1.03-1.13). Significant predictors of cardiovascular mortality included preexisting coronary heart disease, 6.1 (2.8-13.5); macroalbuminuria, 2.5 (1.1-5.8); HbA(1c) level (%), 1.3 (1.1-1.6); and systolic blood pressure (10 mmHg), 1.2 (1.0-1.4). Univariate Cox survival analysis in the normoalbuminuric group revealed that AER above the median of 8 mg/24 h was associated with an increased all-cause mortality risk of 2.7 (0.93-7.69) (P = 0.07). We conclude that abnormally elevated urinary albumin excretion and poor glycemic control indicate a substantially increased all-cause, mainly cardiovascular, mortality risk in NIDDM patients.