In this work, based on real data on the size of the eyeball (in a fetus, in a child, and in young people under 20), we constructed a model function of the growth of the retinal cell tissue.We used this function to construct a theoretical age distribution of retinoblastomas. We constructed theoretical age distributions for four different models of retinoblastoma: a complex mutational model, a third mutational model, a model with a sequence of key events, and a model of a single oncogenic event with two different latencies (hereditary and non-hereditary retinoblastoma).We compared the theoretical age distribution of retinoblastomas with the real age distribution based on SEER data (Surveillance Epidemiology and End Results; program of the American National Cancer Institute). In total, we examined 843 cases in women and 908 cases in men.For all models (separately for women and men), we obtained estimates of the following cancer parameters: the specific frequency of key events (events that trigger cancer); the duration of the latency period of cancer; the number of key events required for cancer to occur.For the composite age distributions, we calculated the theoretical mean age at diagnosis for hereditary and non-hereditary retinoblastomas.The best approximation accuracy (for male and female forms of retinoblastoma) is shown by a model with a sequence of key events.