Supercritical fluid (SCF) dosage is one of the most important parameters for the microcellular injection molding process. Current fierce market competition calls for high part quality and consistency, which places stringent requirements on the accuracy and repeatability of the SCF dosage and the dosing system. In this study, an artificial neural network (ANN)-based SCF dosage control strategy was proposed. An ANN model was constructed and utilized to predict the coming SCF pressure drop in the current injection molding cycle, which is a dominant factor and major source of uncertainty of the SCF dosage. Precompensation to the SCF delivery pressure can then be made according to the prediction to achieve a repeatable SCF pressure drop and thus a repeatable SCF dosage for every injection molding cycle. The result shows that this control strategy can be successfully implemented and that it leads to significant improvements in dosage consistency and part quality. (C) 2011 Wiley Periodicals, Inc. Adv Polym Techn 31: 719, 2012; View this article online at . DOI 10.1002/adv.20230