We present a new approach of predicting the dynamic recrystallization (DRX) of an alloy steel during hot compression testing with an emphasis on higher solution accuracy and practicability than the Johnson-Mehl-Avrami-Kolmogorov (JMAK) model approach. We use not only an accurate closed-form function (CFF) flow stress model which allows to exclude modelling of the strain at 50% recrystallization but also an Avrami kinetics model with improved parameters formulated as the CFFs of state variables; this enhances generality, flexibility, and accuracy. Comparisons of the fitted and experimental flow stresses revealed that accuracy is excellent; the average and maximum errors are less than 2.83% and 4.61%, respectively. The average error of the fitted DRX volume fraction

values in our approach is 5.15% (standard deviation of 2.35%), which is considerably low. The comparison of the grain sizes predicted by the new approach with those by the JMAK model approach also showed that the former is closer to the experimental data than the latter.