The objective of this study was to measure cholesterol concentrations in diabetic patients at the beginning of maintenance hemodialysis treatment and to define their role as predictors of subsequent cardiac death on maintenance hemodialysis. The design of this study consisted of a prospective study of all consecutive diabetic patients newly admitted to 28 German dialysis centers between January 1985 and October 1987. The patients were examined on admission and subsequently followed for 45 months on dialysis. This study included 196 patients, 67 type I (43 male, 24 female, median age 49 years, range 22-73) and 129 type II (54 male, 75 female, aged 64 years, range 37-82). Lipids (total cholesterol, triglycerides low-density lipoprotein (LDL) cholesterol high-density lipoprotein (HDL) cholesterol apolipoprotein B and A and anthropometric indices (body mass index, triceps skinfold thickness) were measured. The outcome was death, i.e., cardiovascular (myocardial infarction, sudden death, other cardiac causes, stroke) and noncardiovascular death during a 45-month follow-up. At the start of treatment, total cholesterol, triglycerides LDL cholesterol, LDL/HDL ratio and apolipoprotein B were significantly higher in diabetics than in healthy controls or patients with standard primary renal disease starting dialysis. Only minor differences were found between males and females and type I and type II diabetics. Fourty-three percent of type I and 50% of type II diabetics died, 61% from cardiovascular causes, mostly myocardial infarction (in 40% reinfarction) and sudden death. On admission, diabetics subsequently dying from myocardial infarction had significantly higher median cholesterol than survivors, i.e., 259 versus 222 mg/dl, and higher LDL cholesterol, LDL/HDL ratio and apolipoprotein B. The conclusion of this study is that high cholesterol is a potent predictor of cardiac death in uremic diabetics treated by maintenance hemodialysis. Controlled trials on the effects of lipid-lowering treatment are desirable.