In the present study, the fatigue behaviour of a clutch diaphragm spring is investigated experimentally and numerically. Differential evolution optimisation algorithm and response surface methodology are used to define optimal variables of the diaphragm springs under constraints both minimum stress and required clamp load. The required clamp load is checked by the chi-square theorem vs. to target clamp load curve. Ten design variables are considered for creating a design of experiment with two steps, including finger shape optimisation with eight variables, load, and stress optimisation with two variables as the aim of maximum fatigue definition. One hundred seventy-five different designs are analysed numerically. As a result of the optimisation study, the optimum design is provided an endless lifetime subject to the required load, minimum stress, and minimum weight. The optimum design is manufactured and tested experimentally.