Men and women patients not previously immunized or tested, attending the Adelaide (South Australia) STD clinic from 1988-1991, were tested for hepatitis B infection, and potential risk factors detected by multiple logistic regression. Of 7055 men and 3425 women patients tested 811 (11.5%) men and 250 (7.3%) women were seropositive. Among men seropositivity was associated with being Asian (odds ratio (OR) = 14.5), being Aboriginal (OR = 2.2), homosexual behaviour (OR = 3.8), intravenous drug use (OR = 3.2) being over 24 (OR = 2.7), previous STD (OR = 1.8), being unemployed (OR = 1.3) and having sex outside the state in the past 3 months (OR = 1.3). Among women seropositivity was associated with being Asian (OR = 10.3), being Aboriginal (OR = 2.4), intravenous drug use (OR = 3.8), being over 24 (OR = 1.6) and having vaginal discharge or dysuria (OR = 1.4). Seropositivity was not independently associated with being a prostitute or having multiple sex partners in the past 3 months. Among seropositive individuals, risk factors were not readily identifiable for 15% of men and 43% of women. Univariate analysis may provide misleading indicators of risk factors because of the confounding influence of other factors, particularly intravenous drug use. In selective vaccination campaigns the target group should be determined on the basis of local circumstances. In South Australia this group should include men with an STD. The success of selective campaigns will be jeopardized by the failure to identify risk factors in many of those who become infected, and in such situations universal vaccination or widespread screening may be more appropriate strategies.