Chemotherapy without radiation has not controlled most intraocular retinoblastoma, perhaps because of the common high expression of multidrug resistance P-glycoprotein that we found in retinoblastoma. Cyclosporin blocks P-glycoprotein-induced efflux of vincristine and teniposide in vitro, and possibly modulates responses to carboplatin. To avoid eye irradiation in bilateral retinoblastoma patients with RB1 germline mutations, which incurs a high second malignancy rate, we added cyclosporin A to a vincristine-teniposide-carboplatin protocol and consolidated chemotherapy responses with focal therapy. We scored patients requiring irradiation, enucleation, or focal ablation of central vision as failures. In 21 study patients, the overall relapse-free rate at a median follow-up of 3.3 years was 76%, with a rate of 92% for newly diagnosed and 50% for previously treated, relapsed retinoblastoma. Our results for the most unfavorable tumors with vitreous seeds (86% at 3.5 years) are better than published success rates of irradiation for similar tumors, or irradiation with the same chemotherapy without cyclosporin (45% at 2. 6 years). These results also exceeded our historic success rate with similar chemotherapy without cyclosporin, focal therapy, and/or radiation in 19 equivalently poor-risk patients (relapse-free rate 37% at a median follow-up of 5.6 years, P = 0.032), 16 of whom were previously untreated (relapse-free rate also 37%, P = 0.012). A better outcome occurred with higher cyclosporin blood levels and projected tissue exposure. Cyclosporin did not enhance the usual chemotoxicity. This clinical study suggests that cyclosporin improves the long-term response of retinoblastoma to chemotherapy, possibly by more than one mechanism.