The history of the application of chemotherapy in the management of retinoblastoma (RB) may conveniently be divided into three eras: initial enthusiasm (from early 1950s to mid 1970s), realism (from late 1970s to mid 1980s) and possibility (from mid 1980s to the future). Available data from each of these eras are reviewed in the clinical categories of: intraocular RB, micrometastatic RB and overt dissemination. The latter is further sub-classified into: orbital invasion, central nervous system involvement and systemic metastases. Experimental models are described with particular emphasis on future directions. Early reports led to initial optimism subsequently dampened by a more critical approach. Recent results with increasingly effective chemotherapeutic regimens offer the possibility of a valid contribution in each of the above clinical settings. A multi-modality approach is recommended optimizing a combination of the most active drugs with continuing refinements of other techniques. In selected patients with intraocular, and particularly bilateral RB, visual outcome may be enhanced by the combined use of non-surgical modalities. Adjuvant treatment of presumed micrometastases needs to be studied within risk categories defined by prognostic factors. Invasion of the ocular coats and/or of the optic nerve are the most relevant factors but there are continued difficulties in defining the extent of involvement and eligibility criteria for such a strategy. Overt dissemination has recently been demonstrated to be curable in each of the three subgroups above. Intensive regimens incorporating cyclophosphamide, vincristine, cisplatinum, etoposide and possibly doxorubicin, plus intrathecal agents in combination with radiation therapy and, in some instances supplemented by bone marrow transplantation have produced promising results. Multi-institutional collaboration has been encouraged by the recently formed International Committee for the Staging and Management of Retinoblastoma, opening the way for prospective clinical trials. At the same time both laboratory and clinical experimental studies are being pursued and may produce further improvements in currently available strategies.