OBJECTIVES: We aimed to assess statistical methods for adjusting survival estimates in the presence of treatment crossover in order to identify which are the most appropriate in a range of scenarios.  Treatment crossover is a common issue in clinical trials of cancer treatments.  Crossover occurs when patients in the control group switch onto the experimental treatment at some point during follow-up.  In such circumstances an intention to treat (ITT) analysis does not address the decision problem faced by health technology assessment bodies, and will result in biased estimates of the overall survival advantage – and therefore the cost-effectiveness – associated with the experimental treatment.  METHODS: We conducted a simulation study to assess the performance of crossover-adjustment methods in a range of scenarios.  We purposefully ran scenarios that did not satisfy the specific assumptions made by the methods, in order to assess their sensitivities.  RESULTS: Randomisation-based methods (eg Rank Preserving Structural Failure Time Models (RPSFTM) and Iterative Parameter Estimation (IPE)) were unbiased only when the treatment effect was not time-dependent.  Observational-based methods (eg Inverse Probability of Censoring Weights (IPCW) and Structural Nested Models (SNMs) with g-estimation) coped better with time-dependent treatment effects but are heavily data reliant, are sensitive to model misspecification and often produced high levels of bias in our simulations.  Observational-based methods are particularly sensitive to the proportion of control group patients that crossover whereas randomisation-based methods are not. CONCLUSIONS: Currently available randomisation-based and observational-based methods for addressing treatment crossover have important limitations.  However, in most circumstances they are likely to lead to lower bias than an ITT analysis, given the decision problem faced in an economic evaluation.  Analysts should consider the treatment crossover mechanism, the control group crossover proportion, the treatment effect associated with different patient groups, and data availability when deciding which method to use to address treatment crossover.