Retinoblastoma, while relatively rare, stands as the most prevalent intraocular cancer. In Switzerland, the survival rate approaches 100%, but it drops to less than 50% in low-income countries. The current treatment options for retinoblastoma rely on a limited set of drugs repurposed from other paediatric cancer therapies. However, concerns about drug toxicity and disease relapse underscore the necessity for new therapeutic approaches. The typical progression of drug candidates from laboratory to clinic involves in vitro validation, which demands precise and clinically relevant in vitro models. Unfortunately, the availability of such models for retinoblastoma is limited, with even fewer suitable for drug discovery endeavours. In the initial section of this thesis, a robust spheroid model optimised for drug screening purposes was engineered. Employing this model, screenings of FDA-approved drugs used in paediatric oncology were conducted, with the potential for repurposing them for retinoblastoma treatment. The findings revealed that receptor tyrosine kinase inhibitors, crizotinib and entrectinib, exhibited efficacy against spheroid cultures, presenting a promising avenue for targeted therapy in retinoblastoma drug discovery. Additionally, drugs displayed the efficacy towards cancer cells did not align with those that exhibited selectivity for cancer cells over normal retinal pigment epithelial cells. This indicates the need for search of potent chemotherapeutics within broader indications and the development of novel molecular entities. In the second part of the thesis, a tumouroid model that accurately replicates key morphological and molecular features of retinoblastoma was established. Moreover, clinical protocols for intravitreal chemotherapy and chemothermotherapy were adapted for use with this tumouroid model. Validation with chemotherapeutic agents currently used in clinical practice resulted in responses akin to those observed in the clinical setting. Lastly, the drugs identified in the first part of the thesis were evaluated using patient-derived tumouroids, developed in the second part of the thesis. The findings revealed that these drugs were effective at high doses but less so at lower doses. Furthermore, there was no single optimal drug for different stages of retinoblastoma. Collectively, the developed retinoblastoma models fit seamlessly into the drug discovery pipeline, aiding in the identification of potential drug candidates tailored to specific clinical regimens. This study underscores the significance of in vitro models that accurately mirror clinical drug responses and emphasises the pressing need for more selective chemotherapeutic options for retinoblastoma.