The band structures of five types of ordered compounds derived from parent zincblende alloys A 1− x B x C and AC 1− x D x have been determined. Included in this study are two novel x= 1 4, 3 4 derived structures, luzonite and famatinite, and three x= 1 2 structures, chalcopyrite and two 1× 1 superlattices oriented along the (0, 0, 1) and (1, 1, 1) directions. The theory combines an empirical tight-binding model for III-V compounds and a valence force-field model of strain. Strain-induced tetragonal and internal distortion as well as the spin-orbit interaction cause a splitting of the top of the valence band. Trends in this splitting and the band-gap variation are studied for the 18 combinations of III-V elements. The Hopfield quasicubic crystal-field model is found to accurately describe this splitting for all chalcopyrite compounds. But this model fails for several (0, 0, 1)-and (1, 1, l)-superlattice compounds containing large strain distortions. The extracted Hopfield crystal-field splitting parameter Δ cf is found to scale linearly with tetragonal distortion for common-anion compounds ABC 2, but follow curvilinearly internal distortion for common-cation compounds. Strain and natural lineup energy modify the band gap significantly from that found in the virtual-crystal approximation for the alloy. For the metastable alloy systems GaAs 1− x Sb x and GaP 1− x Sb x, the experimental bowing of the band gap passes quite close to the results for the band gaps of the seven ordered structures.