Background: Many health care service planning has been based principally on historical utilisation, values and resources. Little attention has been given to determining health service need and planning effective services based on evidence derived from research. As the pressure on resources increases and consumer empowerment develop, health service planning decisions that are made need to be transparent and those who take these decisions will need to be able to produce and describe the evidence on which each decision was based. In an attempt to make the transition from opinion-based to evidence-based decision-making in planning mental health services, the Centre for Mental Health, NSW Department of Health, Australia developed an evidence based population mental health service model determined by mental health need. This population based service model ( the Mental Health Clinical Care and Prevention Model (MHCCP) is a tool to assist systematic considerations of the requirements of comprehensive integrated mental health care and prevention across the lifespan in a population mental health framework. If focuses on the clinical and scientific task of prescribing appropriate 'care-packages' or 'interventions' for individuals and relevant population groups identified through epidemiological mental health prevalence studies.

Methods: Every decision in developing the model was to be based on a systematic appraisal of the best evidence available on: i. Population prevalence surveys; ii. Effective mental health interventions which were appropriate and safe; iii. Evidence-based policy-making iv) Policies on health service financing and organisation. Results Most of the care packages in the service model have been determined by the best available evidence at that time supplemented by clinical opinion. Even in cases for which evidence is difficult to find or poor in quality, evidence was searched for, appraised and presented, even though the decision taken ultimately might have been dominated by values and current utilisation or resources. Unfortunately, general evidence based guidelines are of limited use in prescribing specific quantities of care.

Conclusions: The scope of evidence has to be broadened. Evidence from randomised controlled trials are not readilty available in mental health service planning and information has to be derived from a number of other sources such as gathering evidence about resources and values. The most useful evidence in this planning exercise would have been i. Cochrane style quantitative reviews of key parameters eg. prevalence, incidence, utilisation data and methods of service delivery for specific population groups; ii. Coherent measurement in both population based and service based settings; iii. Quantification of the benefits of interventions in terms of changing client status between service need groups for a defined time period. The Cochrane Collaboration's is well suited for the systematic assembling of service use data and should be broadened to include this.