A new method for correcting a model which simulates the particle size distribution (PSD) has been proposed for a semi-batch emulsion polymerization process involving the copolymerization reaction between vinyl acetate and butyl acrylate. It uses a least squares estimation implemented on the lines of the moving horizon estimation. The work focuses on two important aspects: (i) To see how parameter estimation can be performed without making any assumption on the process or measurement noise. (ii) Using correction factors in the model equations as independent handles for driving the model predictions of the PSD to that of experiments. The approach gives us an insight into the nature of the error presentin the model. The results indicate that the parameter estimation can be made effective by using a minimum number of experimental data for estimation purposes. (C) 2010 Elsevier Ltd. All rights reserved.