A semiclassical theory of asymmetric-top molecule line broadening, developed in part I of this series of papers for linear perturbers, is extended to asymmetric-top perturbers. This model has been successfully applied in part II to the broadening of H2O lines by N2, O2, and Ar. The present applications to the self-broadening of H2O infrared and Raman lines in the 300–900 K range also show the quality of the model. Due to the high electrostatic forces involved in the H2O–H2O interaction, the short-range anisotropic potential has little influence. Nevertheless, the need for a modeling of the trajectory is demonstrated, mainly in the case of high rotational quantum number lines.