Purpose Intraocular retinoblastoma treatments often combine chemotherapy and focal treatments. A first prospective protocol of conservative treatments in our institution showed the efficacy of the use of two courses of chemoreduction with etoposide and carboplatin, followed by chemothermotherapy using carboplatin as a single agent and diode laser. In order to decrease the possible longterm toxicity of chemotherapy due to etoposide, a randomized neoadjuvant phase II protocol was conducted using vincristinecarboplatin vs etoposide-carboplatin. Patients and methods The study was proposed when initial tumor characteristics did not allow front-line local treatments. Patients included in this phase II noncomparative randomized study of neoadjuvant chemotherapy received vincristin-carboplatin (new arm) vs etoposide-carboplatin (our reference arm). They were subsequently treated by local treatments and chemothermotherapy. Primary end point was the need for secondary enucleation or external beam radiotherapy (EBRT) not exceeding 40% at 2 years. Results A total of 65 eyes in 55 children were included in the study (May 2004 to August 2009. Of these, 32 eyes (27 children) were treated in the arm etoposide-carboplatin and 33 eyes (28 children) in the arm vincristin-carboplatin. At 2 years after treatment, 23/33 (69.7%) eyes were treated and salvaged without EBRT or enucleation in the arm vincristin-carboplatin and 26/32 (81.2%) in the arm etoposide-carboplatin. Conclusion Even if the two treatment arms could be considered as sufficiently active according to the study decision rules, neoadjuvant chemotherapy by two cycles of vincristine-carboplatin followed by chemothermotherapy appear to offer less optimal local control than the etoposidecarboplatin combination.