The activation energy, E, of nonisothermal crystallization of polymer melts has been frequently evaluated by the Kissinger equation with positive rates of temperature variation, beta. It is demonstrated that dropping the negative sign for beta is a mathematically invalid procedure that generally makes the Kissinger equation inapplicable to the processes that occur on cooling. Simulations and an experimental examples demonstrate that the use of this invalid procedure may result in erroneous values of E. Alternative methods of estimating the correct E values are suggested.