Objectives: This paper assesses the evidence of changes in sexual behaviour in Zambia, accounting for differences in the composition of survey samples and for evidence of changes in reporting bias. It compares the estimates of sexual behaviour measures obtained using two different survey methodologies.

Methods: Data from five nationally representative household surveys carried out between 1996 and 2003 were analysed for change in selected sexual behaviours.

Results: There is some evidence for an increase in men’s age at first sex. The proportion of people reporting multiple partnerships and not using a condom at last sex has declined since 1996. Unprotected sex with non-cohabiting partners is reported less frequently in the later surveys. The socioeconomic and demographic composition of the survey samples has changed across the years but the declines in behaviour remain statistically significant after adjustment for these changes. There is evidence of a changing reporting bias over the period of interest, with respondents less likely to report a young age at first sex in later surveys.

Conclusions: Between 1996 and 2003, reported sexual behaviour has changed in Zambia among both men and women. Different survey methods produce significantly different estimates of sexual behaviour. This must be taken into account when carrying out trend analyses using data from different sources.