Objectives - To achieve a coherent programme of topics for research in occupational health and safety, with well founded priorities and to relate them to perceived gaps and needs in The Netherlands. Methods - In the first phase of the study 33 key informants were interviewed. In the second phase questionnaires were sent to 150 Dutch experts (including the key informants). Four groups were recruited, originating from: occupational health and safety services; scientific research institutes; governmental and other administrative bodies; and companies. Using the Delphi technique, the experts were asked to prioritise several topics, which were placed under different headings. In the third phase five workshops were organised to elaborate on the highly prioritised topics. Results - The response rates were 86% for the first and 81% for the second questionnaire. In the second round consistency was reached and consensus proved to be satisfactory, so that the Delphi process was stopped. There were surprisingly few differences in opinion between the four groups. The most important heading was 'design/implementation/evaluation of measures', in which the topic cost-benefit analysis of measures had the highest score. 'Assessment of relations between exposure and effect' was the second most important heading. Under this heading, topics on work stress were generally judged to be more important than topics on musculoskeletal disorders, which in turn had higher scores than topics on safety and biological, chemical, and physical hazards. The headings 'occupational rehabilitation/sociomedical guidance' and 'occupational health occupational health services' had about the same priority, closely following the heading 'assessment of relations between exposure and effect'. Conclusions - The general agreement on priorities should provide a sufficiently broad basis for decision makers to initiate a long term programme for occupational health research and development in The Netherlands.