An adaptive stochastic dynamic programming model was used to solve the optimum replacement decision problem for the dairy cow under a range of alternative mastitis control procedures. The model predicted that reducing milk yield losses and somatic cell count penalties by using milking machine test, post-milking teat disinfection and dry cow therapy added approximately £4, £10 and £13 respectively to an original annuity equivalent net present value for the replacement heifer of £286. Assuming that these procedures also reduced involuntary culling due to mastitis by 50% added £8·90 to the annuity. This latter figure indicated that an important part of the benefit of mastitis control procedures might come from a reduction in the culling risk of persistent clinical cases. We concluded that the strength of the dynamic programming model in this context was that it provided an integrated evaluation of the various impacts of each alternative mastitis procedure in the long term, which is essential for correct economic evaluation of mastitis.