This paper discusses stochastic models for the lifelength of non-repairable items under the influence of dominant failure mechanisms like fatigue, corrosion and wear. The Weibull, the lognormal, the inverse Gaussian, and the Birnbaum-Saunders distributions are discussed and compared. Model selection problems based on both field performance data and laboratory data are focused. The consequences of choosing a wrong life model are highlighted. It is concluded that model selection must be based on a thorough knowledge of actual failure mechanisms, and the associated time dependent deterioration. Current knowledge on the relation between some dominant failure mechanisms and the associated life distributions is presented.