Human immunodeficiency virus antigen (HIV-ag) was determined by enzyme immunoassay (EIA) in HIV-antibody (anti-HIV) positive as well as pre-anti-HIV seroconversion sera and the results analysed according to stage of infection, risk group, age and geographic origin. Eleven (19%) of 58 homosexual men tested showed HIV-ag in a serum taken 3–4 months before or one at the time of anti-HIV seroconversion. In another eight (14%) HIV-ag persisted after seroconversion and half of them developed AIDS or AIDS-related complex (ARC) in contrast to none of the other 50 anti-HIV seroconversions. Two (13%) of 16 haemophiliacs tested had HIV-ag only in the first anti-HIV seropositive sample. HIV-ag was present in 86% (30/35) of Dutch homosexual men with AIDS, in 32% (7/22) of men with ARC and in 17% (24/145) of men with persistent generalized lymphadenopathy (PGL) or without symptoms. Three percent (2/60) of sera of asymptomatic i.v. drug users from Amsterdam were HIV-ag positive. Ten percent (1 of 10) of sera from Central Africans with ‘Slim Disease’ were HIV-ag positive. Among infected children from the USA or Europe 89n100% (8/9 and 2/2) of AIDS cases, 67–100% (6/9 and 3/3) of children with ARC and 75% (3/4) of asymptomatic children were HIV-ag positive. The HIV-ag EIA appears to be able to identify HIV infection earlier than the available anti-HIV assays in a significant number of cases. Since persistence of HIV-ag, except possibly in African cases, is strongly associated with clinical deterioration, HIV-ag appears to be a suitable marker for, independent of their clinical status, selecting individuals for antiviral therapy and also for monitoring the efficiency of such therapy.