BACKGROUND: Demands for economic inputs to mental health policy-making, practice decisions and research evaluations have grown considerably in recent years, but the overall supply response has been modest and uneven. AIMS: This paper examines the key historical phases in the development of mental health economics research, and what they imply for the way economics is received and employed. Focusing on the quest for cost-effectiveness, the paper considers challenges for mental health economics. METHODS: An informal review of the growing demand for mental health economics (and how that demand has been expressed), and how economists have responded. RESULTS: Five historical development phases characterize this growth. Initially, the dominant feature is innocence or neglect of scarcity. Cost measures are rarely calculated, cost-effectiveness is not part of the decision-making lexicon and the potential for inefficiency is huge. In the second phase, innocence turns to criticism of attempts to introduce resource rationality, and many clinicians actively reject economics. Health is seen as priceless, and not to be compromised by the pursuit of efficiency. After a period of reluctance there follows impetuosity as the need for economic insights is recognized, but the search for data is desperate and undiscriminating. Poor quality research is conducted, with the risk that decisions are misinformed and perhaps damaging. Once again, resources are inappropriately used. Next follows the constructive development phase: previous mistakes are appreciated and the standards of evaluation improve markedly. Studies are better designed, more likely to be integrated into clinical or policy evaluations, carefully conducted and sensibly interpreted. Inefficiency should be reduced, along with inequity. Finally, there is perhaps a nirvana-like fifth phase in which sophisticated economic studies are widely undertaken, where systematic reviews and meta-analyses help to reveal the wider picture and where findings are readily available to clinicians, managers and providers. Whether such a stage is attainable is open to question. DISCUSSION: Although the number and sophistication of economic evaluations have both increased noticeably over recent years, there remain imbalances. There is little economics evidence on care arrangements or treatments for dementia, most of the neuroses and the disorders of childhood and adolescence. There are many fewer good evaluations of psychological interventions than of drug treatments. Geographically, few economic evaluations are conducted outside Western Europe, North America or Australasia. IMPLICATIONS FOR DECISION-MAKERS AND RESEARCH: Many challenges consequently face the next generation of mental health economics evaluations, both for research economists and for those health care decision-makers who find themselves increasingly having to draw on economics evidence. One challenge is to be fully aware that the information that economists can currently offer may fall short of what decision-makers need. The gap between the two must be fully appreciated. Building more comprehensive pictures of the cost and outcome consequences of different care policies and treatment interventions is one way to bridge this gap. At the same time a sense of perspective must be maintained and promoted. For example, there is growing concern across the world about the high prices of new drugs, yet drug acquisition costs usually represent only a small proportion of total costs. Decisions sometimes appear to be disproportionately focused on small parts of the overall mental health care picture. A similar tendency prompts another challenge, which is to undertake and interpret research so as to overcome, or at least not to exacerbate, the boundary problems that characterize the multi-service, multi-agency reality of many mental health care systems. The adequacy of short-term evaluations - which dominate our field - must be questioned in light of the chronicity of many mental health problems, and of their externality effects (including inter-generational transmission of problems). Although funding will always be a problem, longer-term evidence is needed. So, too, is research that looks at the reasons for inter-individual cost and outcome variations. Economic evaluations should also pay more attention to equity as well as efficiency as a criterion of improved resource allocation. Finally, more economic data should be gathered alongside and not after clinical data, particularly as economic assumptions often appear to drive key practice and policy changes.