A mathematical model of the transmission dynamics of HIV-1 in a heterosexual population stratified by age, sex, and sexual activity (defined by rates of sexual partner acquisition) is presented. The model represents an extension of previous studies with a special focus on patterns of mixing or contact between sexual activity and different age classes of the two sexes. A range of mixing patterns between these groups is specified for both sexes. Mixing is described on two scales from fully assortative to fully disassortative, with random defined either according to numbers of sexual partnerships or numbers of people. The sexual partnerships in the model are balanced by changes in the rates of sexual partner acquisition between particular groups and a range of changes, from only women changing behaviour to only men changing behaviour, are analysed. The pattern of mixing is most influential in determining the shape and magnitude of the epidemic, but the manner in which people choose partners (i.e. dependent on numbers or proportions in the population) is also important. The relative importance of variation in transmission probabilities and mean rates of partner change on the course of the HIV epidemic is also illustrated. The analysis of the sensitivity of predictions to changing parameters in the force of infection term of the model provides a theoretical basis from which the influence of control strategies and the demographic effects of HIV can be analysed.