Evaluating the fatigue damage of marine structures subjected to random environmental loadings is of great practical significance at the design stage. In many situations, the structural responses of marine structures exhibit wide-band non-Gaussian properties. This study presents the fatigue damage assessment of wide-band softening non-Gaussian random processes by a spectral method proposed by Benasciutti and Tovo (2005b). Following this spectral method, the rainflow fatigue damage is approximated by a linear combination of its upper and lower bounds. A refined formulation of the weighting factor is proposed, which depends on the bandwidth parameters, slope parameter of the S‒N curve, and skewness and kurtosis of non-Gaussian processes. By introducing a suitable transformation function, the simulated non-Gaussian process is generated from its underlying Gaussian process. Through case studies, the effectiveness and accuracy of the present method for wide-band non-Gaussian processes are verified by comparisons with the results from the time-domain rainflow counting method and the other existing spectral methods.