The method includes the steps of: storing a plurality of parameter sets; selecting one of the plurality of parameter sets as a test parameter set to be evaluated; measuring performance only for one batch job out of N (N is a positive integer) batch jobs constituting full set performance measurement for the test parameter set; and calculating an evaluation value on the basis of a difference between an integral of measurement values obtained until the performance has been measured for r (r is a positive integer smaller than N) batch jobs by using the test parameter set; and an integral of mean measurement values of the performance for the r batch jobs by using an optimal parameter set which is one of the parameter sets used in the performance evaluation having been performed; determining whether or not the evaluation value has deviated from a predetermined evaluation continuing range; and terminating the evaluation of the test parameter set on condition that it is determined that the evaluation value has deviated form the evaluation continuing range toward performance deterioration. It is preferable that the predetermined evaluation continuing range be of a width from a width W where r is equal to zero, to a width W′ (0≦W′