The method of estimation of kinetic exponent of the Johnson-Mehl-Avrami (JMA) model from non-isothermal DTA or DSC data introduced by Augis and Bennett has been analyzed. It is shown that this method gives reasonably accurate estimations provided that it has been checked by an independent method that the data really correspond to the JMA model. Otherwise, the Augis and Bennett method can lead to wrong interpretation of experimental data. A simple testing method which allows verification of the applicability of this method is proposed. These methods were successfully applied to analysis of non-isothermal crystallization of Ge0.3Sb1.4S2.7 glass.