Postnatal skeletal muscle growth in humans is generally ascribed to enlargement of existing muscle fibres rather than to cellular proliferation. Some evidence of muscle fibre division or splitting was provided in the nineteenth century. This evidence has more recently been supported by fibres obtained from regenerating muscle, and from muscle which has undergone stress-induced growth. Numerous investigators have reported indirect evidence for exercise-induced hypertrophy and hyperplasia. These findings are largely founded on secondary observations of fibre size or number differences expressed relative to muscle cross-sectional area. Since these observations in humans are open to methodological criticism, researchers have developed 3 animal models to represent exercise-induced human muscle growth. These include compensatory hypertrophy, stretch-induced hypertrophy, and weight lifting in trained animals. The results and criticisms of the experiments which have used these models are discussed in this review. In studies of muscle cross-sectional area, errors are created by fibres terminating intrafascicularly. Longitudinal growth of such fibres result in an overestimation of fibre number, and with the use of penniform muscles where fibres do not run parallel to the longitudinal axis of the muscle, the error is compounded. It was concluded that hyperplasia is not yet substantiated, and that new fibres, if present, may be the result of the development of satellite cells. Further experiments are required before a definitive answer can be provided. It is suggested that rigidly controlled exercise studies using contralateral control, fusiform muscles with analysis of individually teased muscle fibres be performed.