The reasons for variation in longevity and disease incidence in untreated control animals in chronic toxicity and carcinogenicity tests in laboratory rodents are reviewed and discussed. Genetic drift, dietary composition, caloric intake, age of animals at the termination of studies, the skill, experience and diligence of the pathologist, the diagnostic criteria used and the appropriateness of statistical procedures are considered to be the most important sources of variation. It is suggested that more attention needs to be paid to the sources of nutrients in animal diets and that 'closed formula' diets should be eschewed. The reliability of historical data depends critically on the extent to which 'like' can be compared with 'like'. The most important variables in relation to histopathological data are considered in this context.