Between 1989 and 1991 a nationwide cross-sectional study of 2837 voluntary subjects was performed in all nine provinces of Zimbabwe to estimate the prevalence of HIV by age, sex and residence (urban/rural). The study areas were 15 randomly selected districts from each of the nine provinces. All collection of blood samples was done at schools, clinics, factories and rural health centres and samples were tested for anti HIV 1 and 2 using both the Abbott Recombinant HIV l/2,3rd generation and the Wellcozyme HIV 1+2. The overall prevalence was 6.9% (95% confidence interval (CI): 6.8-7.0). It was significantly higher in females (8.2%) than in males (5.1%) (P = 0.002) and in urban (11.9%) as compared with rural areas (3.0%) (P=0.0005). Analysing the study as a case-referent study resulted in an odds ratio (OR) of 1.7 for females as compared with males and 4.3 for urban as compared with rural areas. The difference between urban and rural areas remained after standardisation for age, sex and year of data collection but on a lower level (OR=2). The prevalence was higher in provinces where a higher proportion of the subjects was classified as resident in urban areas with one exception (Masvingo province in the southern part of the country (20.8%) and highest an the capital Harare (12.5%). The differences between the provinces remained after standardisation for age, sex and year of data collection. The prevalence of HIV in Zimbabwe was high already in 1989 and increased rapidly between 1989 and 1991, especially in urban areas.