A bulk milk quality assurance programme for Mycobacterium avium subsp. paratuberculosis (Map) in dairy herds was simulated with a stochastic simulation model (JohneSSim). The aim of this study was to evaluate the epidemiological and economic effects of preventive management measures and various test schemes in a simulated population of closed Dutch dairy herds over a 20-year period. Herds were certified as ;low-Map bulk milk' if, with a certain probability, the concentration of Map in bulk milk did not exceed a maximum acceptable concentration of 10(3) Map organisms per litre (based on pasteurisation studies). The programme started with an initial assessment; test-negative herds entered a surveillance procedure and test-positive herds a control procedure. The simulations showed that herd examinations by ELISA for the initial assessment, surveillance and control procedures effectively ensure the quality of ;low-Map bulk milk': > 75% of simulated herds were certified and > 96% of certified herds produced bulk milk with < 10(3) Map/L if the initial herd-level prevalence was 30%. Preventive management measures only had a minor effect on bulk milk quality of certified herds. Culling based on biennial faecal culture was more effective than culling based on annual ELISA. Average total discounted costs for 20-year participation in a programme consisting of initial assessment by ELISA, surveillance by biennial ELISA and control by biennial faecal culture were 16 Euro x 10(3) per herd. In conclusion, this study shows that a bulk milk quality assurance programme for closed Dutch dairy herds is feasible and provides information on the cost-effectiveness of different programmes. The concepts of this study equally apply to other countries because mechanisms of paratuberculosis infection, disease, and testing are comparable in other dairy cattle populations.