In a previous study a model to predict the fatigue S-N behavior of glass fiber reinforced thermoplastics by using a fracture mechanics approach was presented. Using a single flaw size model, some degree of success was observed, particularly for a reinforced polyamide. The model was not successful in predicting the S-N behavior of a reinforced polyester. The earlier study also employed flexural fatigue rather than tensile fatigue data in the calculations because the calculated flaw sizes were more nearly constant as a function of stress level. Subsequently it was shown that the flexural fatigue stress calculations were in error for these types of short glass fiber reinforced plastics owing to the nonlinearity of their stress-strain behavior. In this report we reexamine the utility of the fracture mechanics approach to predict fatigue S-N behavior for both materials using an improved model. Whereas previously a single initial surface flaw was assumed, here we assume multiple flaws growing simultaneously across the sample thickness. The new model is applied to both flexural and tensile fatigue loading. Results demonstrate that this new approach provides accurate predictions of the S-N behavior for both materials under both loading conditions. This reflects the fact that the calculated initial flaw sizes are relatively independent of stress level. No additional adjustable parameters are required if one uses the initial breaking strength of the material as part of the model calculations.