A model was developed to gain insight into the potential clinical and economic impact of antiretroviral therapy for HIV-infected patients. Observed HIV RNA levels and CD4 cell counts are used in the model to estimate the probability that an individual progresses from asymptomatic infection to the first AIDS-defining illness and death and to estimate the total net cost of care and long-term cost-effectiveness of antiretroviral therapy. The model was applied to patients in a clinical trial (Merck protocol 035) that compared the surrogate marker response to triple therapy with indinavir (IDV; 800 mg every 8 hr) plus zidovudine (ZDV; 200 mg every 8 hr) plus lamivudine (3TC; 150 mg twice a day) to double therapy with ZDV+3TC. The model projected that for an individual without AIDS who received triple therapy the progression to AIDS and death would be delayed more than for a patient who received double therapy with ZDV+3TC if no other treatment options were offered. Because of this delay in disease progression, the total discounted cost over the initial 5-year period was projected to be $5100 lower for patients who received triple therapy compared with double therapy if suppression with triple therapy lasts up to 3 years. If suppression with triple therapy lasts up to 5 years, costs were projected to be higher with the triple combination, but 81% of the cost is offset by lower disease costs as a result of fewer patients progressing to AIDS. Over 20 years, total discounted cost was projected to be higher for the triple-therapy regimen primarily because of a longer estimated survival time. At 20 years, the incremental cost per life-year gained by adding IDV to a ZDV+3TC regimen was estimated at $13,229, which is well within the range of other widely accepted medical interventions.