Knowledge about product lifetime derived from real usage (field) data is of great interest in reliability analysis. A preferred source of such knowledge is warranty data, which are generated and updated at no additional cost from customer claims during warranty coverage. However, warranty databases contain only failure-related data, non-failure data is not included. This makes analysis difficult. The present research proposes two alternative methods for estimating product lifetimes from warranty claims data only. Two alternative methods, maximum likelihood and semiparametric, are described for estimating product lifetimes from warranty claims data only. They consider two lifetime variables: exponential failure, which corresponds to random failure mode, and Weibull failure, which corresponds to wear-out failure mode. They use only usage-at-failure data including exponential failure data. Simulation demons frated their applicability.