OBJECTIVES: Recent work has suggested that interventions’ cost-effectiveness should be assessed over their entire lifetime of implementation, not just over the period of use for a single cohort as typically modelled (Hoyle and Anderson, Medical Decision Making, 2010; Hoyle, PharmacoEconomics, 2011). Such lifetime modelling can capture changes in costs and effects over time. These changes in costs and effects can result from price changes, disease dynamics or the application of differential discounting of costs and health effects. METHODS: Suggesting cost-effectiveness be assessed over an intervention’s complete lifetime carries assumptions regarding the nature of the decision problem in healthcare resource allocation. In particular, it suggests resources be allocated on the basis of the total cost-effectiveness over all periods in which it is implemented. This lifetime perspective can conflict with the alternative perspective that resources be allocated on the basis of relative cost-effectiveness within each given period. We discuss a number of simple theoretical examples in which the rank ordering of cost-effectiveness of two interventions is different under the two perspectives. The examples include when the prices of interventions trend and have different expected lifetimes, when differential discounting is applied in certain circumstances, or simply when the price of only one intervention falls following patent expiry. RESULTS: These examples prompt us to consider which perspective is more appropriate. We argue that as health care resource allocation is an ongoing, repeated resource allocation problem, not one over a finite horizon, that the lifetime perspective is not appropriate. CONCLUSION: Advances in decision analytic modelling need to carefully reflect the actual nature of policy choices. The per-period perspective appears more appropriate to healthcare resource allocation problems than the total implementation lifetime perspective. However, the actual resource allocation process is likely to more complex than either perspective alone might suggest.