Retinoblastoma is a rare childhood cancer of the eye. Of the small number of drugs are used to treat retinoblastoma, all have been repurposed from drugs developed for other conditions. In order to find drugs or drug combinations better suited to the improved treatment of retinoblastoma, reliable predictive models are required, which facilitate the challenging transition from in vitro studies to clinical trials. In this review, the research performed to date on the development of 2D and 3D in vitro models for retinoblastoma is presented. Most of this research was undertaken with a view to better biological understanding of retinoblastoma, and we discuss the potential for these models to be applied to drug screening. Future research directions for streamlined drug discovery are considered and evaluated, and many promising avenues identified.