Background: Peritoneal dialysis (PD) and hemodialysis (HD) are often regarded as equivalent choices of renal replacement therapy. However, little is known about the outcome of patients who failed PD and converted to long-term HD. Methods: We reviewed 197 patients who received long-term HD after failed PD in a University hospital from 1994 to 2008 (the PD-first group) and 140 patients who received long-term HD as their initial therapy during that period (the primary-HD group). Their survival rates are compared. Results: The two groups are highly comparable in terms of baseline demographic data. The PD-first group required more temporary dialysis catheters than the primary-HD group (3.1 ± 3.4 vs. 1.5 ± 1.8, p < 0.0001). At 5 years, the actuarial survival of the PD-first group was significantly lower than that of the primary-HD group (39.9 vs. 59.7%, p < 0.0001), while technique survival was similar (30.4 vs. 30.1%, p = 0.7). When analysis on actuarial survival was performed for patients who survived the first 12 months on HD, the 5-year survival became similar (65.2 vs. 68.8%, p = 0.5). During the first 12 months on HD, independent predictors of actuarial survival of the PD-first group were duration of PD, Charlson’s comorbidity score, type of permanent access and serum albumin before conversion; after 12 months, independent predictors of actuarial survival were Charlson’s comorbidity score, total Kt/V, residual renal function, fat-free edema-free body mass before conversion and baseline peritoneal transport. Conclusion: Patients who were converted to long-term HD after failed PD had a higher mortality than patients who used HD as the primary modality of renal replacement therapy. The excessive mortality, however, was limited to the first 12 months after conversion, and the technique survival was similar between the two groups. Vascular access is a common problem in patients who failed PD.