In an attempt to verify some of the current conflicting results concerning the impact of relevant prognostic factors in the retinoblastoma therapy, the authors took into consideration, for statistical analysis, the series of 459 cases included in the Italian Registry for retrospective study of retinoblastoma. Although this series appears large enough, problems related to the continuously changing approaches to the disease and the consequent lack of standardization often make it difficult to draw significant conclusions. Hence, while historical (retrospective) analysis often allows the manipulation of a great number of data, particularly in the case of relatively rare diseases, prospective randomized controlled trials are strongly recommended to standardize definitely the relevant prognostic criteria. These and other problems related to retrospective analysis are discussed in detail.