Objectives: To determine the prevalence of HIV and risk behaviour in a national sample of homosexual men in the Netherlands. To adjust this prevalence estimate for selection factors connected with a blood test. Methods: Participants in the study including a blood test, (n=308, also 147 particpants without a blood test) were recruited in 1991/1992 among 1134 male respondents of a non-HIV specific postal inquiry without a blood test in 1989 among readers of the "Gay Krant" a national magazine for homosexuals. There are no indications that readers of the Gay Krant have a different risk of HIV infection than other homosexual men. On base of self-reported serostatus in 1989 the measured seroprevalence in 1991/1992 is adjusted in a logistic model for selection through participation in the blood test. Results: The prevalence among the 308 participants in the blood test was 6.5% (95% CI 4.0-10.0). Participation in the blood test was not significantly related to self-reported risk behaviour in 1989 or self-reported serostatus in 1989. The prevalence of HIV in 1991/1992 among the 826 other respondents of 1989 is estimated by the model to be 5.2%. The HIV prevalence in 1991/1992 among male readers of the Gay Krant resident in the Netherlands is estimated to be 5.6%. Using a statistical simulation technique the 95% confidence interval of this prevalence is estimated to be 3.4-7.8%. The estimated number of HIV infected homosexual men in the Netherlands in 1991/1992 lies between 6500 and 15000. Of the participants 40% had anal sex with steady partners and 25% with casual partners in the last 6 months. Of those 79% (with steady partners) respectively 42% (with casual partners) did not or not always use condoms. In all 38% had unprotected anal sex in the last 6 months. Risk behavior with steady partners increased between 1990 and 1991/1992. Conclusions: The spread of HIV among homosexual men in the Netherlands is substantial, even outside of Amsterdam. The level of risk behavior is high and shows an increase. It is possible to carry out seroepidemiological research among homosexual men without participation being selective on HIV status.<br>