Estimates of cost-effectiveness analyses are typically obtained either directly from 'trial' based analyses or indirectly via surrogate endpoints in 'model' based analyses. Data from clinical trials that include both surrogate and final endpoints can be used in a joint analysis that combines these two approaches. This joint approach allows the inclusion of information regarding the effects of treatment on surrogate endpoints while relaxing the strong assumption of 'conditional independence' associated with indirect model-based analyses. An example cost-effectiveness analysis of Chronic Disease Self-Management Programme is used to compare the different approaches. It is shown that despite using a common data set, the different analytic approaches produce differing estimates of the cost-effectiveness of the intervention and the value of future research. The paper concludes by discussing the selection of the appropriate analytic approach. Copyright © 2011 John Wiley & Sons, Ltd.