A prototype model of the mean radius flow path of a four-stage,high speed 1 MWe axial steam turbine was optimized by using evolution algorithms,DE(differential evolution) algorithm in this case.Also the cost-benefits of the optimization were inspected.The optimization was successfully performed but the accuracy of the optimization was slightly less than hoped when compared to the control modeling executed with the CFD(computational fluid dynamics).The mentioned inaccuracy could have been hardly avoided because of problems with an initial presumption involving semi-empiric calculations and of the uncertainty concerning the absolute areas of qualification of the functions.This kind of algebraic modeling was essential for the success of the optimization because e.g.CFD-calculation could not have been done on each step of the optimization.During the optimization some problems occurred with the adequacy of the computer capacity and with finding a suitable solution that would keep the algorithms within mathematically allowable boundaries but would not restrict the progress of the optimization too much.The rest of the problems were due to the novelty of the application and problems with preciseness when handling the areas of qualification of the functions.Although the accuracy of the optimization results was not exactly in accordance with the objective,they did have a favorable effect on the designing of the turbine.The optimization executed with the help of the DE-algorithm got at least about 3.5 % more power out of the turbine which means about 150 000 € cost-benefit per turbine in the form of additional electricity capacity. A prototype model of the mean radius flow path of a four-stage, high speed 1 MWe axial steam turbine was optimized by using evolution algorithms, DE (differential evolution) algorithm in this case. Also the cost-benefits of the optimization were inspected. The optimization was successfully performed but the accuracy of the optimization was slightly less than hoped when compared to the control modeling executed with the CFD (computational fluid dynamics). The mentioned inaccuracy could have been difficult because of problems with an initial presumption involving semi- empiric calculations and of the uncertainty concerning the absolute areas of qualification of the functions. This kind of algebraic modeling was essential for the success of the optimization because egCFD-calculation could not have been done on each step of the optimization. problems occurred with the adequacy of the computer capacity and with finding a suitable solution that would keep the algori thms within mathematically allowable boundaries but would not restrict the progress of the optimization too much. The rest of the problems were due to the novelty of the application and problems with preciseness when handling the areas of qualification of the functions .Although the accuracy of the optimization results did not exactly in accordance with the objective, they did have a favorable effect on the designing of the turbine. The optimization executed with the help of the DE-algorithm got at least about 3.5% more power out of the turbine which means about 150 000 € cost-benefit per turbine in the form of additional electricity capacity.