Two‐constant Kubelka—Munk (KM) theory with experimentally fitted values accounts quite well for the total reflectance of opaque mixtures of paints, but systematic errors indicate the need for better theory. No simple satisfactory alternative exists, however. Errors from the Saunderson correction for surface reflections are comparable to errors from KM theory, and opinions differ on the best values for the diffuse internal reflection coefficient. Since the surface corrections can be applied separately from diffuser theory, an extra variable is available. Even for isotropic scattering, theory shows that the internal surface reflectance ranges from 0.57 when the diffuser reflectance is high to 0.65 when it is low, while the corresponding Saunderson coefficient ranges from 0.52 to 0.65. For anisotropic scattering, the coefficient is virtually independent of the phase function for single scatter when the reflectance is high, but critically dependent when it is low. Values of around 0.4 are predicted for the absorption band, unless the particles are very sm