The slow pyrolysis of four biomass samples was studied by thermogravimetry (TGA) at different heating rates. The samples belonged to different botanical classes/genera, and their mineral matter content showed a high variation. A distributed activation energy model (DAEM) was used due to the complexity of the biomass samples of agricultural origin. The common features of their decomposition kinetics were sought by evaluating 12 experiments of four biomasses simultaneously by the method of least-squares. Two parallel DAEM reactions with a Gaussian distribution of the activation energies were sufficient for an acceptable fit between the experimental and simulated data. Common means and deviations of the activation energies were required for all four samples. The reactivity differences between the samples were expressed by the differences between the preexponential factors, while the weights of the parallel reactions described further differences between the samples. Altogether, 20 unknown model parameters were estimated from 12 experiments. When the method of least-squares was used on the mass loss rate (DTG) curves, one of the obtained partial curves showed a sharp peak with a small variation of E. This was associated with the cellulose decomposition. The other partial curve had a much wider E distribution and was assumed to include the decomposition of hemicellulose, lignin, and extractives. The evaluation of the sample mass (TGA) data resulted in wider partial peaks than the ones obtained from the DTG data. To exclude the possibility of any mathematical artifact, the evaluation was also carried out on the analytical integrals of the DTG curves.