The complexity of the phenomena which arise during the heating of the various substances seldom can be described by a single reaction kinetic equation. As a consequence, sophisticated models with several unknown parameters have to be developed. The determination of the unknown parameters and the validation of the models requires the simultaneous evaluation of whole series of experiments. We can accept a model and its parameters if, and only if we get a reasonable fit to several experiments carried out at different experimental conditions. In the field of the thermal analysis the method of least squares alone seldom can select abest model or abest set of parameter values. Nevertheless, the careful evaluation of the experiments may help in the discerning between various chemical or physical assumptions by the quality of the corresponding fit between the experimental and the simulated date. The problem is illustrated by the thermal de-composition of cellulose under various experimental conditions.