An analysis of Hall measurements on n‐type and p‐type GaP is performed by means of least‐squares fits including the temperature dependences of the thermal impurity activation energies which result from the theoretically proposed linear dependence ED on N D+1/3. By comparison of fit results obtained with the standard model (ED = const.) and with the ED(n)‐model, it is shown that the neglect of the ED(n) dependence causes especially too high values for N c′/g and NA. Therefore the replacement of the standard model by the ED(n)‐model yields lower values for the density of states effective masses of GaP (mc* = 1.09m0 and mV* = 0.52m0,) and higher values for the slope α of the regression line in the ED − N A1/3 plot. Using the ED(n)‐fit it is possible to determine for a single sample the optical value of the activation energy of the dominating impurity which allows conditionally the identification of the chemical nature of this impurity.