In this article, the flexural fatigue behavior of wood flour filled polypropylene (PP) composite samples was investigated, and a probabilistic approach is presented to evaluate the fatigue reliability. A nondimensional analysis was adopted to establish a prediction equation for the fatigue life of the composites, and the S-N curves and survivability diagrams are then constructed to evaluate the fatigue reliability for the wood plastic composites. The predicted results in the nondimensional fatigue model are better related to the testing data than the general S-N curve, and they could be used as the predictive model for fatigue life analysis of the wood plastic composites. The best-fit S-N curve and its corresponding bounds based on 95% confidence are provided, and the fatigue data are well distributed within a 95% confidence range. A two-parameter Weibull distribution function is used to analyze statistically the fatigue life test results of composite samples based on the probability of survival concept which is proposed to incorporate the probability of survival and the fatigue life time of the composites at a specified lifetime. Weibull graphics are plotted for the composite samples used in the fatigue test, and the SN curves are then drawn for different reliability index (RI) of 0.01, 0.1, 0.5, 0.9, 0.99, and 0.368 using these data. These S-N curves are introduced for the identification of the first failure time as reliability and safety limits. It is shown that the nondimensional fatigue model and two-parameter probability distribution model can be used successfully for reliability evaluation and lifetime estimation of wood plastic composites. POLYM. COMPOS., 31:553-560, 2010.