It is unclear whether peritoneal dialysis (PD) compared with hemodialysis (HD) confers a survival advantage in end-stage renal disease (ESRD) patients with coronary artery disease (CAD). This hypothesis was tested in a national cohort of 107,922 patients starting dialysis therapy between May 1, 1995, and July 31, 1997. Data on patient characteristics were obtained from the Center for Medicare and Medicaid Services Medical Evidence Form (CMS) and linked to mortality data from the United States Renal Data System (USRDS). Patients were classified on the basis of CAD presence and followed until death or the end of 2 yr. Nonproportional Cox regression models estimated the relative risk (RR) of death for patients with and without CAD by dialysis modality using primarily the intent-to-treat but also the as-treated approach. Diabetic patients (DM) and nondiabetic patients (non-DM) were analyzed separately. Among DM, patients with CAD treated with PD had a 23% higher RR (95% CI, 1.12 to 1.34) compared with similar HD patients, whereas patients without CAD receiving PD had a 17% higher RR (CI, 1.08 to 1.26) compared with HD. Among non-DM, patients with CAD treated with PD had a 20% higher RR (CI. 1.10 to 1.32) compared with HD patients, whereas patients without CAD had similar survival on PD or HD (RR = 0.99; CI, 0.93 to 1.05). The mortality risk for new ESRD patients with CAD differed by treatment modality. In both DM and non-DM, patients with CAD treated with PD had significantly poorer survival compared with HD. Whether differences in solute clearance and/or cardiac risk profiles between PD and HD may explain these findings deserves further investigation.