OBJECTIVES: The National Institute for Health and Care Excellence recommends utilising external data to evaluate the validity of extrapolation beyond trial follow-up. The objective was to demonstrate the use of external data to guide long-term survival predictions where only short-term trial data are available. METHODS: Four-year patient-level data for chlorambucil and ofatumumab+chlorambucil from the COMPLEMENT-1 trial were available; the survival rate was over 70% at four years.  Published 18-year Kaplan-Meier data for chlorambucil from the C9011 study, which compared fludarabine and chlorambucil, were used to simulate a patient-level dataset. These data were not used directly as outcomes have improved substantially since patients were enrolled (1990-1994) and the chlorambucil dosing regimens differed.  However the curve was used to guide the extrapolation of the COMPLEMENT-1 data using a three-stage approach: 1) Survival functions were fitted for the COMPLEMENT-1 and C9011 chlorambucil arms with an indicator for study. 2) The average treatment effect (chlorambucil versus ofatumumab+chlorambucil) was estimated from survival analysis for both arms of COMPLEMENT-1 and the C9011 chlorambucil arm, with indicators for treatment and study. 3) For ofatumumab+chlorambucil, long-term survival was estimated by applying the treatment effect from stage 2 to the function for chlorambucil from stage 1. Exponential, gamma, Weibull, log-normal, log-logistic and Gompertz functions were fitted. RESULTS: Weibull, Gompertz, and gamma functions provided reasonable fits during trial follow-up and plausible extrapolations (assessed by Akaike information criterion, Bayesian information criterion, graphical diagnostics, and expert clinical opinion). Survival predictions were markedly lower than conventional functions fitted to COMPLEMENT-1 data only (E.g. Weibull predictions for chlorambucil at 20 years were 8% vs 26%, respectively). CONCLUSIONS: Using long-term external data to guide the extrapolation provided plausible predictions for chlorambucil upon which alternative scenarios for the continuation of treatment effect observed in COMPLEMENT-1 could be explored.