Patients infected with HIV, including those with AIDS-related complex and AIDS, and failing treatment with antiretroviral agents such as zidovudine, have been evaluated following addition of trichosanthin to the antiretroviral agent regimen. This ribosomal inhibitory protein is specifically cytotoxic for HIV-infected macrophages and lymphocytes. Ninety-three patients were treated with trichosanthin, using a schedule of weekly, then monthly, intravenous injections of 1.2 mg of drug in combination with antiretroviral agents, usually zidovudine. Side effects included myalgias, fevers, mild elevation in liver function tests, and mild-moderate anaphylactic reactions, which respond well to therapy with steroids and/or benedryl. Reversible mental status changes were noted in two patients, both receiving concomitant therapy with ddI. Clinical responses to trichosanthin treatment were monitored primarily by changes in laboratory parameters, particularly levels of CD4+ T lymphocytes. In the total population evaluated for efficacy (85 patients) there was a significant increase in CD4+ cell levels after initiation of trichosanthin therapy. A second analysis performed on 72 patients measured the rate of change of CD4+ cells during therapy, using an "area under the curve" analysis. During therapy there was a median increase of 1.2 cells/mm3/month. In patients in the top 25th percentile, this increase was greater than 8.4 cells/mm3/month. In 59 of the 72 patients, responses could also be monitored by comparing the rate of loss of CD4+ cell levels on antiretroviral agents (zidovudine or ddI) alone, during the year prior to initiation of trichosanthin, to the rate of change when trichosanthin was added to the treatment regimen. During the period before trichosanthin treatment (311 +/- 11.7 days) the median loss of CD4+ cells was 6.91 cells/mm3/month. Addition of trichosanthin to the treatment regimen resulted in a median gain of 1.1 CD4+ cells/mm3/month.