Overall survival (OS) is traditionally modelled in economic evaluations of anti-cancer drugs.  However, OS is commonly associated with problems such as immaturity of the data, or confounding due to treatment switching or use of inappropriate treatments after progression. Fortunately, analysis of historical trials reveals that there is good evidence across a range of cancers that the mean time in post-progression survival (PPS) is equal between treatment arms, i.e. ΔPPS = 0.  Therefore, we recommend that the default position is to assume equal mean times post-progression.  If there is no a priori biological reason to suppose that the PPS times are likely to differ between treatments (e.g. due to differences in cross-resistance or long term toxicities between treatments), our recommendation is that it should be assumed that the mean time in progressive disease is equal between treatment arms if any of the following apply: OS is very immature; treatments post-progression are substantially imbalanced between treatment arms; in particular, treatment switching has occurred at progression; treatments post-progression are different to those routinely given in clinical practice; only single arm trials are available.  If none of the above apply, or if there are a priorireasons to suggest that ΔPPS differs from 0, then the recommendation is to model OS and PFS in the traditional way. For chronic cancers, it is recommended that analyses should either assume equal times post initial treatment or equal time post progression. The assumption that ΔPPS = 0 substantially simplifies the economic analysis because cost-effectiveness becomes insensitive to OS.  The methodology has been endorsed twice by NICE appraisal committees in assessments of drugs for chronic myeloid leukaemia. The cost-effectiveness of several drugs recently assessed by NICE are re-calculated using the methods proposed.  Next, we give simplified formulae for the maximum drug price acceptable for reimbursement under the methodology.