Prediction of prognosis and response to treatment is of utmost importance for patients with kidney cancer. The complicated natural course of kidney cancer causes some difficulties in the prediction of prognosis. Prognostic factors for kidney cancer can be grouped as tumor related, patient related, or molecular factors. Several prediction models have been developed by using these prognostic factors. These models can be for either localized or metastatic disease. Those for localised disease can be pre- or post- nephrectomy models. In recent years molecular markers have been integrated into the existing models. Integration of molecular markers may allow the prediction of response to targeted treatments in the future.