The genetic basis of retinoblastoma is reviewed and the following conclusions are drawn: 1) The mode of inheritance of the hereditary variety of retinoblastoma (R) is autosomal dominant with about 90% penetrance. 2) About 68% of inherited cases are bilateral, and about 32%, unilateral. There is an intrafamilial correlation between penetrance as measured by segregation ratio and expressivity as measured by the fraction of bilaterally affected patients. 3) The vast majority of all R patients are sporadic cases, i.e., they are the only affected members of otherwise unaffected families. The porportion of bilateral cases is much lower among sporadic than among hereditary cases. 4) All bilaterally affected patients with sporadic R and patients with unilateral sporadic R with more than one primary tumor have to be regarded as germ cell mutants; they will transmit the gene to 50% of their offspring. Only 10%-12% of unilateral sporadic cases are germ cell mutants; 88%-90% are nonhereditary; in these cases the tumor is probably caused by a somatic mutation. 5) In a minority of cases, deletion of the chromosome segment 13q14(=intersitital deletion of the long arm of chromosome 13) has been observed. In addition to R, the patients show a variable degree of general or mental retardation; often there are few external indications of a chromosome aberration. Other chromosome studies suggest anomalies of chromosome 13 in tumor tissue even in cases not showing an anomaly of this chromosome in blood cultures, and possibly a slightly increased chromosome instability. 6) Patients with bilateral, and possibly in general with hereditary, R run an increased risk of becoming affected with other tumor diseases, such as osseous sarcomas, in later life. 7) Knudson's hypothesis of two mutational steps leading to both the hereditary and the nonhereditary variants of R is discussed critically, and the alternative possibility is suggested that in the nonhereditary variant a single mutational step--possibly a small chromosome aberration--could be enough to produce a tumor. 8) Evidence indicating a possible viral origin of R is cited, and animal experiments are mentioned in which R-like tumors have successfully been produced by local DNA virus inoculation. 9) As a consequence of improved survival and reproduction of R patients, an increased in the incidence of R and in the proportion of bilateral cases among all R patients must be anticipated. 10) Detailed rules for genetic counseling in families affected by R are given.