To critically analyze the literature linking microalbuminuria with total and cardiovascular mortality and cardiovascular morbidity in non-insulin-dependent diabetes mellitus (NIDDM) and to quantify the risk. A combination of retrieval techniques (MEDLINE, SCISEARCH, and handsearching published bibliographies) was used to find all relevant articles based on title and abstract and "Methods" sections. Unpublished data on albumin excretion rate were sought from large NIDDM cohort studies. A total of 264 citations were retrieved, of which 11 cohort studies were selected for inclusion in the overview, representing a total of 2138 patients followed up for a mean of 6.4 years. Patient age was similar across cohorts. Duration of NIDDM ranged from newly diagnosed to 13 years. The prevalence of microalbuminuria ranged from 20% to 36% in the 8 cohorts that excluded patients with clinical proteinuria. All studies reported either a trend or a significant association between microalbuminuria and total mortality or cardiovascular morbidity or mortality; the overall odds ratio for death was 2.4 (95% confidence interval, 1.8-3.1) and for cardiovascular morbidity or mortality, 2.0 (95% confidence interval, 1.4-2.7). We found no evidence of reporting bias. Microalbuminuria is a strong predictor of total and cardiovascular mortality and cardiovascular morbidity in patients with NIDDM.