Eleven subjects who presented with a clinical illness characteristic of primary HIV-1 infection were treated with 1 g zidovudine daily for a median period of 56 days (range, 28–111 days). Primary HIV-1 infection was confirmed in each subject by seroconversion and virus isolation. The acute phase of the illness resolved a median of 4 days (range, 3–14 days) from commencement of zidovudine. Six subjects reported symptoms that may have been side-effects of zidovudine, the most common being nausea in four subjects and headache in two. Treatment was discontinued in one subject who had persistent headache and nausea. Haemoglobin, haematocrit and erythrocyte counts decreased and mean corpuscular volume increased significantly during the treatment. None of the subjects developed anaemia and none required dose modification or blood transfusion as a result of haematological side-effects. There were no significant differences in the granulocyte count or the lymphocyte count during any week of treatment when compared with baseline levels. There were no significant differences in T-cell subset numbers of the subjects during treatment compared with a group of historical controls. HIV-1 was isolated from several subjects during and after termination of zidovudine treatment. The results of this investigation indicate that zidovudine is a safe drug to administer to people with primary HIV-1 infection. There was no clear evidence, however, of any clinical benefit in terms of resolution of the acute illness and no indication that the treatment would prevent development of persistent infection. Nevertheless, the results urge the establishment of a placebo-controlled trial to further evaluate this treatment and its effect on long-term outcome in people with primary HIV-1 infection.