OBJECTIVE To determine if HIV-1 RNA and CD4 lymphocyte thresholds for the initiation of highly active antiretroviral therapy (HAART) are associated with clinical response to therapy. METHODS Observational cohort study. METHODS Johns Hopkins Hospital HIV Clinic. METHODS HIV-infected adults. METHODS Patients initiating HAART (n = 530) were compared with concurrent patients who did not receive HAART (n = 484). METHODS Progression to a new AIDS-defining illness or death. RESULTS The average duration of follow-up for the cohort was 22 months. HAART resulted in decreased disease progression among persons with fewer than, but not more than, 200 x 10(6) CD4 lymphocytes/l prior to treatment. Among persons receiving HAART, plasma HIV-1 RNA level prior to therapy was not associated with HIV disease progression within CD4 T-lymphocyte count strata. In a Cox multivariate proportional hazards model that adjusted for age, sex, race, prior opportunistic infection, and CD4 T lymphocytes, < or = 200 x 10(6) CD4 lymphocytes/l was the strongest predictor of disease progression. HIV-1 RNA level prior to starting HAART of < 5000 copies/ml, 5001-55 000 copies/ml, or > 55 000 copies/ml was not associated with disease progression on therapy, particularly among persons with > 200 x 10(6) CD4 lymphocytes/l. There was no sex difference in disease progression on treatment. CONCLUSIONS Our data suggest that current guidelines for initiating HAART should place greater emphasis on CD4 lymphocyte than HIV-1 RNA level for both men and women. Further longitudinal follow-up will be needed to better ascertain whether HAART initiated at > 200 x 10(6) CD4 lymphocytes/l is effective in slowing disease progression.