OBJECTIVE: To evaluate a support program for patients co-infected with the human immunodeficiency virus and tuberculosis in terms of its impact on clinical outcomes and resource utilization. METHODS: We compared co-infected patients receiving Community-Based Accompaniment with Supervised Antiretrovirals (CASA) with matched patients receiving standard of care (control group) in two health districts of Lima, Peru. We recorded clinical outcomes, costs of the intervention, and health care utilization by each patient during 24 months of follow-up. RESULTS: There were 33 patients in each group, representing 58.0 person-years (py) in the CASA group and 45.6 py in the control group. At 24 months of follow-up, the CASA group had a lower hazard of dying or defaulting from treatment (HR adj 0.34, 95% CI 0.12–0.98), experienced fewer hospital days (IRR adj 0.37, 95% CI 0.14–0.99) and had fewer out-patient visits (IRR adj 0.75, 95% CI 0.63–0.89). Assigning costs to significantly different measures of health care utilization using WHO-CHOICE (World Health Organization–Choosing interventions that are cost effective) data, CASA was associated with savings of US $551/py. Considering intervention costs of US $2097/py, the net costs of CASA were US $1546/py. CONCLUSIONS: Our intervention was associated with clinical improvements and reduced health care utilization, which significantly offset the cost of the intervention over 2 years of follow-up.