In recent years there has developed a growing recog nition that publicly financed health services need to utilise explicit principles of resource allocation.1 Policy makers have come to recognise this need in the light of evidence which suggests that increases in the supply of health services will be matched by an increase in the demand that is made upon them. Thus increasing expenditure on medical resources does not of itself solve the persistent problem of scarcity; resources will always be insufficient for the demands upon them. Combine this fact of scarcity with the acknowledgement that there are relatively efficient and inefficient ways of providing and delivering health services, and the need for explicit principles of allocation becomes plain. Yet the recognition of the need for principles does not of itself help us to develop such principles. The question still remains: on what basis should one allocate the limited resources that are available to health services in order to bring about an ethically acceptable allocation ? A particular example of this dilemma is provided by the problem of striking the correct balance between preventive and curative medicine. Thus, it has recently become fashionable to assert that available resources might be used to best effect if they were devoted more extensively to preventive measures with the implication that they might be switched from the curative services.2 But the notion of best effect is left undefined in this context. What is meant by the assertion that preventive services are likely to be more efficient than curative services ? Efficiency in this context only has meaning in relation to a specified set of objectives. So in order to judge the claim that one use of resources is more efficient than another, we need to formulate a set of objectives in terms of which we can measure the performance, or output, of different types of services, or inputs. In order to formulate these objectives it is neces sary that the community have a method for placing a value on the lives of its members. The manner in which the decision on allocating health services is made is then best pictured as follows.3 The com munity decides, by means of the political process, how much money is to be devoted to health services in total. This decision represents, in effect, the community's evaluation of health services by com parison with other desirable goods, like transport facilities, education, defence and housing. Having made this decision, which in practice is usually little more than an incremental decision from previous years' expenditures, the public authorities must decide how best to use available resources. In the short term they may not have much room for manoeuvre. Personnel will have been trained, capital goods reacquired and revenue thereby com mitted. But in the longer term the public authorities should be seeking to move towards the most efficient use of resources possible, otherwise effort will be wasted and things left undone which ought to be done. What criterion should the public authorities use in making this decision on allocation ? A seductively attractive, if ambiguous, proposal in this context is that, in their health expenditures, the public authorities should seek to maximise the value of lives saved or prolonged. This decision-making process in practice begs the question of what value the community assigns to the lives of its members. What typically emerges from the process is the goal simply of maximising the number of lives saved or the average length of life for members of a given population. Thus the widespread use of population mortality rates or standardised mortality ratios in the appraisal of health service performance provides a measure of effectiveness only on the assumption that what is of interest is simply the number of lives saved or the numbers multiplied by the increases in average length of life.4 No differential value is attached to the lives of different persons. Each is counted for one and no one for more than one. The criterion has in fact a certain ethical attractiveness to it. Because lives are weighted equally there will be no deliberate bias or discrimination involved in the selection or treatment of beneficiaries from the health services. Moreover, provided the technical problems can be solved in estimating the effectiveness of various forms of service, that is provided what economists