Background: Many studies have compared patient survival outcome between hemodialysis (HD) and peritoneal dialysis (PD); however, time-varying risks of dialysis modality have been rarely investigated. This study aimed to investigate dialysis modality switch and its association with the survival outcome in the Korean population. Methods: Data from the Korean Society of Nephrology were used. A total of 21,840 incident dialysis patients who started dialysis in or after 2000 were analyzed. For the survival analysis, both proportional and non-proportional hazard assumptions were applied. For the modality switch, time-varying covariate Cox regression was applied. Results: During the median follow-up of 8 years, PD group showed increased adjusted hazard ratio (HR) of 1.248 (95% CI 1.071–1.454, p = 0.004) for mortality. Interaction of PD status with female sex was significant with an HR of 1.080 (95% CI 1.000–1.165, p = 0.050). Dialysis modality switch was associated with increased HR of 1.094 (95% CI 1.015–1.180, p = 0.019), albeit switch from PD to HD did not show significant HR until 6 years. Interestingly, time-varying risk analysis showed a decreased HR of PD after 10 years in the non-switcher group, which was consistent in patients with high traditional risk factors (with diabetes, elderly). Conclusions: PD was associated with increased HR of mortality in the first 8 years, then it was associated with decreased HR of mortality after 10 years. Dialysis modality switch was associated with increased mortality risk, but switch from PD to HD within 6 years did not show significant hazard of mortality.