A linear regression equation has recently been developed to predict the activation energy of static recrystallisation for hot-worked austenite based on the testing of more than 40 steel grades. The equation has been shown to be fairly accurate for a wide variety of chemical compositions. However, certain steel grades, such as interstitial-free steels and steels with relatively high contents of Nb, Ti or Si, behave as outliers with respect to the model predictions. The reasons for the discrepancies have been unclear. As another approach, it is assumed here that the activation energy can be estimated from the drag effect exerted on moving grain boundaries by one or two, the most efficient, solute atoms, rather than the sum effect from all solute elements as suggested by the regression equation. It turns out that the behaviour of the steels can be rationalised relatively well in this way without outlier behaviour. However, due to limited data, the new approach is less accurate for ordinary steel grades than that based on the simple regression equation.