This paper studies the characteristics of a proposed reliability-dependent hazard rate function for composites under fatigue loading. The hazard rate function, in terms of reliability R, is in the form of e+c (1-R)p called (ecp) model, where e denotes the imbedded defects of material strength, c the coefficient of strength degradation, and p the memory characteristics of distributions of both applied stress and fatigue strength during the cumulative damage process. By taking a typical residual strength model in Monte Carlo simulation, this paper presents the time changing of the residual strength distribution and hazard rate of composite under various constant-amplitude cyclic stresses. The values of (e, c, p) are decided by fitting hazard rate function to the data generated in simulation. The results show that, under a suitable suggested value of e, p is a constant depending on the characteristics of stress distribution as well as the residual strength model used in Monte Carlo stimulation, and c is correlated to the maximum cyclic stress in a power-law relationship. Only by knowing the initial strength distribution and the maximum cyclic stress, the fatigue life can be easily estimated by integrating the reliability with time or its equivalent, i.e., the reciprocal of hazard rate function with reliability. Finally, by a proposed approximated equation of fatigue life, the (ecp) model is checked to be highly consistent with S-N curve in both the physical means and the equation form. The analysis presented here may be helpful in designing and maintenance planning of composite under fatigue loading.