A group of 709 randomly selected males aged 18-55 years living in the city of Amsterdam, the Netherlands, were questioned about sexual lifestyle and related behavior. A substantial proportion (10%) of these men appeared to be homo- or bisexual. Of these homo- and bisexual men, 16.5% were predicted to be HIV seropositive, using their behavioral information in logistic regression models derived from a second study among 506 homo- or bisexual volunteers who were questioned and tested for HIV antibodies as well. Generalizing the results of the sample, the magnitude of the homo- and bisexual population in Amsterdam is estimated to be 21,500 (95% confidence interval: 17,000-26,000) of whom 3500 are HIV seropositive. This is in agreement with estimations on the basis of AIDS surveillance data. The study shows that in the absence of serological data it is possible to estimate HIV prevalence, given that the relation between the presence of HIV antibodies and sexual activity or indicators for sexual activity is known. Since serological data are lacking in many occasions, this method may be of help to estimate HIV prevalence in other geographic locations and risk groups.