OBJECTIVES: Immunotherapies such as ipilimumab and IL-2 show delayed but durable response leading to stabilization of symptoms and extended OS after an initial drop-off in the KM curve.  Our objective was to review and challenge currently available economic modelling methods when applied to such emerging therapies with new mechanisms of action (MoA). METHODS: As alternatives to standard OS extrapolation methods which fit ‘traditional’ parametric survival distributions to patient-level data, two different methods were explored in the modelling of OS beyond the trial duration (55 months) for the novel immunotherapy ipilimumab. In the first approach, the hazard rate from the Kaplan-Meier (KM) curve between 24 and 36 months (before reaching a plateau) was used to extend the curve. In the second approach, different parametric curves were fitted to the period of 18 months onwards. Akaike’s Information Criterion (AIC) was used to determine the best fit curve. RESULTS: When compared to standard OS extrapolation methods, both methods exhibited a better visual fit to the data. Both approaches allow the hazard of the extrapolated tail to be based on a section of the KM curve that is more appropriate in describing the long-term survival of these patients. The hazard rate approach does not allow for a formal comparison with AIC, but allows extrapolation in line with the clinical interpretation. The ‘parametric curves’ approach allows for a statistically better fit with the patient level data using conventional AIC criteria. Both methods are in line with long-term observations of immunotherapy. CONCLUSION: For novel cancer therapies whose KM curves are not well described by standard survival distributions, other methods of extrapolation should be explored in conjunction with an understanding of the clinical rationale. In this case study, two alternatives are presented that describe the OS of immunotherapy patients in a more suitable way.