PURPOSE: Following curative treatment for localised renal cell carcinoma (RCC), up to 30% of patients develop tumour recurrence. Prognostic scores are essential to guide individualised surveillance protocols, patient counselling and potentially in the future to guide adjuvant therapy. In metastatic RCC, prognostic scores are routinely used for treatment selection in clinical practice as well as in all major trials. METHODS: We performed a literature review on the current evidence based on prognostic factors and models for localised and metastatic RCC. RESULTS: A number of prognostic factors have been identified, of which tumour node metastasis classification remains the most important. Multiple prognostic models and nomograms have been developed for localised disease, based on a combination of tumour stage, grade, subtype, clinical features, and performance status. However, there is poor level of evidence for their routine use. Prognostic scores for patients with metastatic RCC receiving targeted treatments are used routinely, but have limited accuracy. Molecular markers can improve the accuracy of established prognostic models, but frequently lack external, independent validation. CONCLUSION: Several factors and models predict prognosis of localised and metastatic RCC. They represent valuable tools to provide estimates of clinically important endpoints, but their accuracy should be improved further. Validation of molecular markers is a future research priority.