We sought to determine whether lower mortality rates reported with hemodialysis (HD) at home compared to hemodialysis in dialysis centers (center HD) could be explained by patient selection. Data are from the United States Renal Data System (USRDS) Special Study Of Case Mix Severity, a random national sample of 4,892 patients who started renal replacement therapy in 1986 to 1987. Intent-to-treat analyses compared mortality between home HD (N = 70) and center HD patients (N = 3,102) using the Cox proportional hazards model. Home HD patients were younger and had a lower frequency of comorbid conditions. The unadjusted relative risk (RR) of death for home HD patients compared to center HD was 0.37 (P < 0.001). The RR adjusted for age, sex, race and diabetes, was 44% lower in home HD patients (RR = 0.56, P = 0.02). When additionally adjusted for comorbid conditions, this RR increased marginally (RR = 0.58, P = 0.03). A different analysis using national USRDS data from 1986/7 and without comorbid adjustment showed patients with training for self care hemodialysis at home or in a center (N = 418) had a lower mortality risk (RR = 0.78, P = 0.001) than center HD patients (N = 43,122). Statistical adjustment for comorbid conditions in addition to age, sex, race, and diabetes explains only a small amount of the lower mortality with home HD.