Empirical models for grain growth predictions are briefly reviewed. It is demonstrated that such models are inadequate for quantitative prediction of austenite grain growth during reheating of as-cast microstructures in microalloyed steels. This inadequacy is mainly attributable to the inability of empirical models to account for abnormal grain growth. Basic principles of grain growth are therefore revisited in an attempt to develop a mathematical model which can account for abnormal grain growth. Such a model has been developed for Ti : N balanced microalloyed steels and is presented here. It is shown that quantitative predictions of austenite grain growth generated from this model fit well with experimental grain growth data obtained during reheating of as-cast slabs of Ti, Ti + Nb as well as Ti + Nb+Mo containing microalloyed steels.