Summary Objective To evaluate the efficacy of voluntary counselling and testing (VCT) for HIV/AIDS in changing risky sexual behaviour in central Mozambique. Method Longitudinal cohort study of men and women aged at least 18 years from October 2002 to June 2003. We interviewed 622 participants in VCT groups and 598 in non-VCT groups. The interviews occurred before counselling and 4 and 6 months afterwards. Results Reported use of condoms while having sex with a friends/prostitute increased over each time period in the VCT group and between baseline and first visit in the non-VCT group. Both men and women in the VCT group increased their condom use over time, but the women in the non-VCT group did not. Reported always/sometimes use of condoms for both literate and illiterate subjects was higher and rose over time in the VCT group. Conclusion People who undergo voluntary counselling and testing fro HIV/AIDS change their behaviour, presumably as a result of their counselling.