In polyolefin processes the melt index (MI) is the most important control variable indicating product quality. Because of the difficulty in the on-line measurement of MI, a large number of MI estimation and correlation methods have been proposed. In this work, we propose a new MI predicting model in which the reaction heat term in existing prediction models is replaced by temperature. The proposed model is compared with other prediction models as well as operating data to evaluate MI predicting performance. The present model shows least predicting errors compared to existing models.