This paper aims to enable the relevant use of water main service lifetime and failure data to build a medium or long term infrastructure management plan. Firstly, how to estimate the service lifetime distribution of water mains using observations of decommissioning times which are possibly left-truncated and predominantly right-censored, is shown. Three methods are presented: a non-parametric method another based on the parametric Weibull distribution, and a third based on the parametric Herz distribution. An application with actual data related to grey cast iron water mains of two large French and German water distribution networks illustrates the implementation of the theoretical methods. The paper then investigates the link between failure rate and pipe renewal, and discusses the use of observation-based service time survival functions for infrastructure asset management.