In order to study the relative impact of a steady homosexual relationship on the presence of HIV antibodies, a cross-sectional study was undertaken of 719 homosexual men in Oslo during the period of 1983-1987. 74/719 (10%) were HIV positive. 37/74 (50%) of the HIV positives and 185/645 (29%) of the HIV negatives were having a steady homosexual relationship that had lasted six months or more (chi-square test, p < 0.05). In a multivariate logistic regression analysis the adjusted odds ratio of being HIV positive was 2.3 (1.3-4.3, 95% confidence interval (CI)) for homosexual relationship lasting 1/2-4 years and 3.1 (1.5-6.6, 95% CI) for homosexual relationship lasting 5 years or more as compared with not being in a homosexual relationship. The number of male partners and frequent anal intercourse were also significantly associated with presence of HIV antibodies. The results suggest that situational variables associated with a steady partnership may be barriers to HIV-preventive measures.