Uniform melt front velocity is recommended for injection mold filling to minimize the part nonuniformity. A method of profiling the injection velocity for such a uniform mold filling is presented in this paper. Based on a neural network model developed for estimating the melt flow length from online measurable variables, the profiling problem is transformed into an optimization to minimize the difference between the predicted melt flow length and a given ramp. The rate of the ramp determines the rate of the melt front velocity traveling in the cavity during filling. Experiments with different molds show that the proposed method is effective in profiling the screw injection velocity to achieve a uniform mold filling. (C) 2006 Wiley Periodicals, Inc.