The paper provides a brief review of the significance of heterogeneity in sexual behaviour to the transmission dynamics of HIV and the spread of AIDS. It addresses the formulation of mathematical models to encapsulate variability in the rate of sexual partner change, the structure of networks of sexual partner contact, and age dependency in sexual activity. A particular focus is the significance of high or low preference (or choice) of sexual partners from an individual's own sexual activity class (defined on the basis of partner change rate or other criteria). Numerical studies of model behaviour reveal that a high degree of assortativeness (high level of "like with like" mixing) results in a more rapid initial spread of HIV, a smaller overall epidemic, and the possibility of a multipeak or long and drawn out epidemic, by comparison with those induced by high degrees of disassortativeness (low level of "like with like" mixing) in sexual contact patterns. The structure of a sexual network is shown to be a major determinant of the temporal pattern and magnitude of an epidemic. The demographic impact of HIV in developing countries is also examined in the context of choice matrices based on age as well as sexual activity. The potential demographic impact of AIDS is shown to be enhanced by age dependency in levels of activity (high in the young and lower in older age classes), a male preference for females of younger age, and a higher efficiency of HIV transmission from male to female than vice versa. The paper ends with a discussion of the need for better quantitative data on sexual behaviour.