Survival analysis applying proportional hazards models was used to investigate genetic and non-genetic factors affecting ewe longevity and lamb survival of sheep in Ethiopia. Data were obtained from an on-station closed nucleus breeding program of Menz sheep and an on-farm Awassi×Menz sheep crossbreeding project. A total of 695 ewes and 1890 lambs born from the nucleus population of Menz sheep were used for the analysis of ewe productive life and lamb survival to yearling age respectively. In addition, 5530 lamb records of purebred local and crossbreds with proportions of ~25–50% Awassi, collected from three locations were used for the analysis of lamb survival from birth to weaning age. The effects of year, ewe parity and litter weight at weaning were significant (p<0.05) for ewe productive life. On-station lamb survival to yearling was affected by year and breeding value of yearling weight of the lamb, indicating that faster growing animals had substantially higher survival rates. Animal model heritability estimates for lamb mortality ranged from 0.02 to 0.10. While there was a strong genetic trend for growth rate, as evidenced by estimated breeding values for different birth years, estimated breeding values for lamb survival were variable across years with decreasing trend. These results indicate no antagonism between growth rate and survival under conditions prevailing in the nucleus system. Routine genetic evaluation for survival is suggested. Under farmer conditions, the effects of year, season, sex and location effects were significant (p<0.05) on lamb survival to weaning age whereas breed composition (local versus crossbred) of dam as well as of the lamb were not significant. This seems to indicate that crossbreeding of local animals with Awassi sheep does not have an adverse effect on the survival of lambs under farmer conditions typical for the Ethiopian highland regions included in this study.