While it is generally accepted that non-insulin-dependent diabetes mellitus (NIDDM) increases atherosclerotic risk, controversy remains as to whether this effect is greater for women than men (thus reducing the usual gender differential). Furthermore, it is unclear to what extent changes in general risk factors may account for this increased risk. The literature was reviewed with meta-analyses. Gender specific overall relative risks with 95% confidence intervals for coronary heart disease (CHD) mortality (ICD codes 410-414) were calculated. Similarly, overall gender specific odds ratios for prevalent myocardial infarction (MI) are presented. Data are generated from both fixed effects and random effects models. Frequency counts of studies showing specific cardiovascular disease (CVD) risk factor effects in diabetes are given as is the number of studies showing diabetes to be an independent risk factor. The overall relative risk (the ratio of men to women) for CHD mortality in diabetes was 1.46 (1.21-1.75) and 2.29 (2.05-2.55) in nondiabetes suggesting that the gender differential is reduced in diabetes. However, heterogeneity was high (P < 0.001). Exclusion of studies that were exclusively in elderly subjects eliminated heterogeneity (P > 0.05), but retained a separation of the confidence intervals. Overall odds ratios (men:women) show no suggestion of a diabetes effect on the gender difference for prevalent MI, 1.77 (diabetes) and 1.79 (no diabetes). The effects of six general CVD risk factors were unclear, although the largest study showed a clear effect of cholesterol, smoking, and blood pressure. All 10 studies in women report diabetes to be an independent risk factor as do 8 out of 12 studies in men. NIDDM reduces the gender differential in CHD mortality, but not for prevalent MI (or other end points). Although the effect of specific CVD risk factors is inconsistent across studies, this is likely to reflect limited sample size and power. The major three risk factors, cholesterol, blood pressure, and smoking, probably operate in NIDDM but do not fully explain the increased risk of CVD in NIDDM.