The objective of this paper is to get detailed information on the dynamics of condom use with various partner types, including consistency of use, and to estimate the frequency of unprotected intercourse. The study is based on a prospective cohort study of 222 urban workers in urban Zimbabwe. Respondents' sexual behaviour and condom use were tracked for a period of six weeks using daily interviewer-administered questionnaires. In total, the observation period covers 9324 person-days, during which 4601 sexual contacts were reported. The results indicate that most sexual encounters with casual partners are protected, and condom use with these partners is very consistent. However, although the majority of sex acts with regular non-marital partners are protected, only four out of ten men with regular partners used condoms consistently. Because many men have both regular and casual partners, fewer than one in seven males used condoms in all their sex acts. We estimate that nearly half of all males have more than 85 unprotected sex acts per year. Although many of these acts are with spouses, who have comparatively low risk, we estimate that about one in five men have 25 or more unprotected acts with regular non-marital partners annually. WI conclude that regular partners are likely to be a major source of HIV transmission in Zimbabwe, partly because condom use is inconsistent and partly because they account for a large share of all sex acts. Given that over one in four adults in the general population are HIV-positive, and that some studies show that married persons have the highest HIV prevalence, it is important for future HIV programmes to focus on increasing levels of protection with regular partners.