► The original Dirlik model is modified to estimate the high cycle fatigue damage. ► The model combine the frequency based fatigue and the temperature effect. ► The modified model enables computationally fast fatigue damage estimation. This paper proposes a temperature modified Dirlik method to estimate the high cycle fatigue damage for uniaxial loadings caused by random vibrations directly from a power spectral analysis. Besides, the methodology for combining the frequency based fatigue analysis with the temperature effect is represented. This approach is based on a new definition of loading as a random Gaussian process. The fatigue damage estimation of the high pressure die-cast aluminium alloy AlSi 9Cu 3 is investigated at elevated temperatures. Finally, numerical simulations on the known power spectral densities with different shapes at different temperatures are performed in order to establish proper dependence between the temperature modified Dirlik method, the rainflow cycle counting, the linear cumulative fatigue damage and the spectral bandwidth parameters. The proposed method enables computationally fast fatigue damage estimation for the random loadings and the temperature histories.