We propose a numerical method for predicting the distribution of mechanical properties, such as elastic modulus and yield stress, which appear in an injection molded part. In this method, the mechanical properties are directly predicted from the fiber orientation, the thermal history and the flow one of fiber-reinforced thermoplastics experienced during an injection molding process. Multivariate polynomial equations are derived with a neural network method, as prediction equations. Predicted mechanical properties agree well with measured ones for specimens cut out from several locations in a molded plate of fiber-reinforced polypropylene.