Epidemiological and observational studies are needed in nephrology for evidence-based medical decision and global knowledge of renal patients. Using strong methodology, such studies are useful to formulate hypotheses for further explanatory studies or clinical trials. Survival analysis of dialysis patients are based on the usual and robust Kaplan-Meier and Cox regression methods. Nevertheless, their use should take into account the specificities of the dialysis population, especially when non-constant risks for death with time and sub-groups analyses are considered. In addition, survival curves from birth or standardised mortality ratio are able to provide a new view of survival by changing of analytical perspective. Our aim is to summarize the specificities of survival study methodology in dialysis patients using concrete examples in French cohorts.