The effect of changes in sexual behaviour of a population stratified into high and low sexual activity classes on the temporal trends of HIV/AIDS is evaluated, using a simple mathematical model (Anderson et al. 1989), to obtain the activity class on which control programmes should be targeted. The initial population parameters are such that the introduction of infection into the population results in epidemics in both activity classes (Uche and Anderson 1993a): particular focus is on a population with initial mean rate of partner change of the high and low sexual activity classes given as 200 and 2.5 respectively and transmission probability 0.1. The simple deterministic model, with constant recruitment rate of susceptibles into the population, is used to examine individually the effect of varying the mean rate of partner change of the high activity class from 200 to 10, the mean rate of partner change of the low activity class from 3.0 to 1.75 and the transmission probability from 0.5. to 0.025. The proportion of the total recruitment rate into the high activity class is in all cases fixed at 1/11.

Assuming fully assortative mixing, two peaks in seroprevalence of infection were realised. It was observed that changes in the mean rate of partner change of the high activity class from 200 to 10 reduced the the magnitude of the first peak by 2.6% from 8.7% and delayed the time of its occurance by only 12.4 years with no effect on the magnitude (difference of about 0.4%) or the time of occurance of the second peak.