Important decisions are frequently made from the analysis of product life and field data using various graphical and analytical methods. Simple methods, such as fitting a Weibull distribution to the data, however, may give wrong answers, especially when estimates beyond the data range are desired, under certain conditions. The first part of this paper reviews the major purposes of product life analyses, briefly describes some of the standard methods and indicates some pitfalls associated with the use of simple methods. The hazards are illustrated by the analysis of human mortality and electromechanical device data. The second part of the paper discusses pitfalls resulting from product mixtures, proposes some general models for product life analysis and suggests remedial approaches.