Despite the enormous cardiovascular disease epidemic among maintenance hemodialysis (MHD) patients, total hypercholesterolemia seems paradoxically to be associated with better survival. It was hypothesized that similar paradoxic associations also exist for serum LDL, HDL, and triglycerides. A 3-yr (July 2001 through June 2004) cohort of 15,859 MHD patients was studied in the United States from DaVita dialysis clinics where lipid profile was measured in at least 50% of all outpatients during a given calendar quarter. Cox proportional hazard models were adjusted for case mix and surrogates of malnutrition-inflammation complex. Both total and LDL hypercholesterolemia showed a paradoxic association with better survival. Hypertriglyceridemia (>200 mg/dl) also showed a similar trend, but serum HDL cholesterol did not have any clear association with survival. The association between a low serum LDL <70 mg/dl, which was prevalent among almost 50% of all MHD patients, and a higher all-cause death risk was robust to multivariate adjustment. In the subgroup analyses, these paradoxic associations persisted among most subgroups, although they tended to be stronger among hypoalbuminemic (<3.8 mg/dl) patients and those with a lower dietary protein intake (<1 g/kg per d). However, in black patients, a high serum LDL (>100 mg/ml) was associated with adjusted cardiovascular death hazard ratio of 1.94 (95% confidence interval 1.12 to 2.38; P = 0.02). Despite inverse associations between hyperlipidemia and survival, black MHD patients with high LDL show almost two-fold increase in cardiovascular death risk. Although these associations may not be causal, they call into question whether specific subgroups of dialysis patients are better targets for cholesterol-lowering therapy.