OBJECTIVES: To develop a method of economic evaluation and triage for research prioritisation, before the funding decision. DATA SOURCES: Existing models were researched focusing on MEDLINE, HealthSTAR, IBSS and HEED. REVIEW METHODS: Papers of primary relevance that included a proposed model were reviewed in detail, and their models appraised using criteria adapted from the EUR-ASSESS project and the authors' previous experience. From this the PATHS model was developed. It assumes three or more possible alternative outcomes or scenarios in terms of research results: 'favourable' to the technology being assessed, 'unfavourable' or 'inconclusive'. An associated flow of benefits or disbenefits, costs or savings is identified for each potential research outcome depending on the likely implementation of the results as judged by experts. These benefits and costs are weighted and discounted in the model to give an expected incremental cost-effectiveness ratio (EICER). EICERS could be estimated for any number of research areas or proposals to inform funding prioritisation. The model was tested and evaluated on three case studies identified in liaison with the NHS R&D HTA programme and the UK Medical Research Council. These case studies were funded research projects, where full evaluation was underway and where results would be reported during the PATHS project. The studies were selected to include surgery or other invasive procedures, and non-invasive health services projects (a fourth case study did not complete during the course of the study). The three case studies included randomised controlled trials of early surgery or observation for small abdominal aortic aneurysms, infusion protocols for adult pre-hospital care, and postnatal midwifery support. RESULTS: Each of the three assessments indicated net clinical benefit or no clinical loss of benefit, in addition to health service cost savings in excess of the cost of the trial. For two case studies, the value of the proposed trial, as evaluated by the model in the prediction, was consistent with the ex post evaluation, thus providing positive tests of the value of the model. In the third case meaningful ex post analysis was not possible as very poor compliance with the trial protocol (indicated in the ex ante evaluation) seriously undermined its conclusions. During the study, at the request of the UK HTA programme, the model was also applied to a funding request for a large randomised trial of beta-interferon for multiple sclerosis treatment. CONCLUSION: The PATHS model has a useful part to play in the research prioritisation process. Its strengths lie in its emphasis on the impact of research results on policy and practice (the keystone for NHS research) and net effects on health benefits and costs. It assesses the cost-effectiveness of the research and may identify ways to enhance the research design, endpoints relevant to implementation, analytical methods and dissemination. Further research is recommended to investigate the scope for synthesising the strengths of the PATHS model with other approaches including value of information; to compare ex ante and immediate ex post assessments of implementation with long term follow-up of actual implementation; and to assess the robustness of such approaches to the choice and number of experts used.