Aims: To test the applicability of an extended version of the theory of planned behaviour for the study of condom use intentions among large samples of young people in South Africa and Tanzania. Methods: Baseline data of a randomized controlled trial of school-based HIV/AIDS prevention programmes were used. The setting comprised secondary schools in the regions of Cape Town, Polokwane and Dar es Salaam. Participants were 15,782 secondary school students. The main measures were scales for intentions, knowledge, risk perceptions, attitudes, perceived social norms and perceived self-efficacy regarding condom use. Results: Seven variables accounted for 77% of the variance in intentions to use condoms: attitudes ( ¼ 0.17), injunctive norms ( ¼ 0.27), self-efficacy ( ¼ 0.41), gender (lower condom use intentions among females), being a student at the Dar es Salaam site (lower scores than students in Cape Town and Polokwane), socioeconomic status (higher intentions with higher status), and access to condoms (higher intentions with higher access). Conclusions: Our results are comparable to those of studies conducted in Europe and the USA. Social cognition models such as the theory of planned behaviour are applicable in understanding the correlates of condom use intentions in African contexts.