The concept of variable activation energy in solid-state reaction kinetics has caused considerable debate. Activation energy variation has been detected by isoconversional or "model-free" calculation methods, which generate activation energy as a function of reaction progress. The relationship between calculation methods and artifactual variation in activation energy was investigated in this work by employing model-fitting and isoconversional methods to analyze both simulated and experimental nonisothermal data. The experimental data was for nonisothermal sulfameter-dioxolane solvate desolvation by TGA. We show that variable activation energy in simple reactions could be an artifact resulting from the incorrect application of isoconversional methods. (c) 2005 Elsevier B.V. All fights reserved.