OBJECTIVES:  To assess long-term survival outcomes for nivolumab in renal cell carcinoma (RCC) predicted by three model structures: a partitioned survival model (PSM) and two variations of a semi-Markov model (SMM). METHODS:  Three economic model structures were developed and populated using parametric curves fit to patient-level data from the CheckMate 025 trial, a phase 3 study comparing nivolumab with everolimus in previously treated patients with RCC. All models consisted of three health states: progression-free survival (PFS), progressed disease, and death. The PSM estimates stated occupancy using an “area under the curve” approach from overall survival (OS) and PFS curves, whereas the SMMs explicitly derived transition probabilities to calculate patient flow between health states. One SMM assumed that post-progression survival (PPS) was independent of PFS duration (PPS Markov); the second SMM assumed differences in PPS based on PFS duration (PFS-PPS Markov). RESULTS:  In CheckMate 025, the 2-year OS rate was 51.7%; the PSM, PPS Markov, and PFS-PPS Markov predicted 2-year OS rates of 52.9%, 55.0%, and 54.2%, respectively. OS curves derived in the PSM provided the closest fit to the trial data. The five year survival, conditional upon surviving one year, was 18.2%, 10.0%, and 22.1% for the PSM, PPS Markov, and PFS-PPS Markov, respectively. The mean OS for nivolumab estimated by the PSM, PPS Markov, and PPS-PFS Markov was 41.1, 36.2, and 45.2 months, respectively. CONCLUSIONS:  All three model structures provided a good fit to the trial data, but different long-term survival was predicted by each model. This would likely lead to differences in estimated cost-effectiveness results if used in economic evaluations, which has particular implications for health technology assessment bodies. In the absence of long-term survival follow-up data, the use of external datasets and clinical opinion to validate survival predictions become important to justify choice of model structure.