Experimental observations of creep life in structural systems exhibit significant scatter. As a result, probabilistic methods that incorporate the associated uncertainties in residual life assessment methodologies have been developed. Most studies in the literature adopt Gaussian models for the noise. However, Monte Carlo simulations with such models lead to possibilities of physically unrealistic negative damage growth increments in certain sample damage growth realizations. This study investigates the use of alternative models for noise in damage growth analyses and compares these predictions with those obtained when Gaussian models are used. A continuum damage mechanics based approach is used for obtaining the thermal creep damage growth in a nuclear power plant component. Numerical results are presented to highlight the salient features arising from this study. © 2014 ASM International.