The process by which Canadian provinces decide what to cover, or to remove from coverage, under Medicare often seems ad hoc, without clear guiding principles. It has been argued, most notably in the context of the Oregon experiment, that cost-effectiveness analysis could be used as an explicit tool to determine what services should or should not be covered. However, while cost-effectiveness analysis has been widely accepted as the "best" available alternative in health care decision making, unlike cost-benefit analysis, it is not well rooted in welfare economic theory. Hence, many questions remain as to whether this type of framework actually yields the best outcome for a society. Furthermore, it is not clear from the underlying principles of cost-effectiveness analysis if distributional issues are really avoided. This paper considers that argument in a welfare economics framework, looking at the methodology used to elicit preference orderings. It is well known that cost-effectiveness analysis requires strong assumptions about the utility function, if it is to fit into the framework of welfare economics. Furthermore, there are distributional issues when aggregating individual utility functions, which have been the centre of some controversy. With reference to the Oregon Project, eliciting individual utilities through quality of life measures such as rating scales and aggregating these into a social welfare function may not lead to the desired outcome, which was apparent in the problems that arose from Oregon's first prioritized list. A better method is to elicit a social welfare function from individuals, using the person trade-off method. This incorporates the societal perspective and avoids difficulties associated with aggregating individual utilities. Prioritizing health services in this manner will lead to a socially acceptable outcome, because it elicits the social welfare function that reflects society's preferences about the prioritization of health services to be covered.