METHODS We performed a randomized controlled clinical trial to assess the safety and efficacy of a 1 microgram/h ganciclovir implant for the treatment of newly diagnosed cytomegalovirus (CMV) retinitis in patients with the acquired immunodeficiency syndrome (AIDS). Patients with previously untreated peripheral CMV retinitis were randomly assigned either to immediate treatment with the ganciclovir implant or to deferred treatment. Standardized fundus photographs were taken at 2-week intervals and analyzed in a masked fashion. The study end point was progression of retinitis based on the photographic assessment. RESULTS Twenty-six patients (30 eyes) were enrolled. The median time to progression of retinitis was 15 days in the deferred treatment group (n = 16) vs 226 days in the immediate treatment group (n = 14) (P < .00001, log-rank test). During the study, 39 primary implants and 12 exchange implants were placed in immediate-treatment eyes, deferred-treatment eyes that progressed, or contralateral eyes that developed CMV retinitis. Postoperative complications in the total series included seven late retinal detachments and one retinal tear without detachment. Final visual acuity was 20/25 or better in 34 of 39 eyes. The estimated risk of developing CMV retinitis in the fellow eye was 50% at 6 months. Biopsy-proven visceral CMV disease developed in eight (31%) of 26 patients. The median survival was 295 days. CONCLUSIONS The ganciclovir implant is effective for the treatment of CMV retinitis. Patients with unilateral CMV retinitis treated with the implant are likely to develop CMV retinitis in the fellow eye, and some patients will develop visceral CMV disease.