This paper considers how the information provided by economic evaluation for decision-makers can fail to optimise use of health resources and how assessment of the relative cost-effectiveness of health care interventions can be misleading unless heterogeneity within populations is taken into account. The cost-effectiveness of an intervention is not a point estimate but an average chosen from within a distribution of different results. The normal interpretation of the distribution around that point is often mistakenly assumed to be the 'white noise' of measurement error. In reality this variance is a combination of measurement error and true heterogeneity of results. There remains an overemphasis on pursuing certainty which stems from the fact that the methods involved were originally devised to measure dichotomous outcomes not continuous ones such as cost-effectiveness ratios. It is argued in this paper that more consideration be given to the heterogeneous nature of costs and effects across populations in analysis and policy making.