HIV-1 prevalence typically rises more rapidly at young ages in women than in men in sub-Saharan Africa. Greater susceptibility to infection on exposure in women is believed to be a contributory factor as is greater exposure to previously infected sexual partners of the opposite sex. We investigated the latter hypothesis using data from a field study in rural Manicaland, Zimbabwe. Quantitative data on onset and degree of sexual activity, numbers of partners, concurrent partnerships, condom use, and partner characteristics were used in conjunction with epidemiological data on age and sex specific prevalence of HIV infection to do statistical analyses of association between key variables. Mathematical models and qualitative data were used to aid analysis and interpretation. Older age of sexual partner was associated with increased risk of HIV-1 infection in men (odds ratio 1.13 [95% CI 1.02-1.25]) and women (1.04 [1.01-1.07]). Young women form partnerships with men 5-10 years older than themselves, whereas young men have relationships with women of a similar age or slightly younger. Greater number of lifetime partners is also associated with increased risk of HIV (1.03 [1.00-1.05]). Young men report more partners than do women but infrequent coital acts and greater use of condoms. These behaviour patterns are underpinned by cultural factors including the expectation that women should marry earlier than men. A strong gender effect remains after factors that affect exposure to infected partners are controlled for (6.04 [1.49-24.47]). The substantial age difference between female and male sexual partners in Manicaland is the major behavioural determinant of the more rapid rise in HIV prevalence in young women than in men. Theoretical studies have suggested that this difference is an important determinant of observed epidemiological patterns but the study reported in this paper provides clear empirical evidence of association.