The main aim of the present study was to examine the economic consequences of a reduction in the incidence of clinical mastitis (CM) at herd level under current Swedish farming conditions. A second objective was to ask whether the estimated cost of CM alters depending upon whether the model reflects the fact that in different stages of lactation, CM gives rise to different yield-loss patterns or postulates just one type of yield-loss pattern irrespective of when, during lactation, CM occurs. A dynamic and stochastic simulation model, SimHerd, was used to study the effects of CM in a herd with 150 cows (9000 kg of energy-corrected milk per cow-year). Four herd types, defined by production level and reproductive performance, were modelled to investigate possible interactions between herd type and response to a reduction in the risk of CM. Technical and economic results, given the initial incidence of CM (25.6 per 100 cow-years), were studied together with the consequences of reducing the initial risk of CM by 50% and 90% throughout lactation and the consequences of reducing the initial risk by 50% and 90% before peak yield. A conventional way of modelling yield losses – i.e. one employing a single yield-loss pattern irrespective of when, during the lactation period, the cow develops CM – was compared with a new modelling strategy in which CM was assumed to affect production differently depending on its lactational timing. The effect of the choice of reference level when estimating yield losses was investigated by comparing the results obtained using the potential yield of mastitic cows, had they not developed CM, with those obtained using the yield of non-mastitic cows. The yearly maximum avoidable cost of CM at herd level was estimated at €14 504, corresponding to 6.9% of the net return given the initial incidence of CM. Expressed per cow-year, the maximum avoidable cost was €97. The cost per case of CM was estimated at €428. Herd types all responded in a similar manner to the reduced relative risk of CM. There were no major differences in the results obtained using the new and the conventional modelling strategy, with the exception of the cost per case of CM. Similarities between the results obtained using the two methods were particularly evident when the mastitic cows’ own yield level, had they not developed CM, was used as the reference for production in healthy cows when yield losses were estimated. It was concluded that the conventional way of modelling yield losses is adequate and should, for the foreseeable future, be used in decision support systems.