In this work, we present the first ab initio study for Cr(NH3)6(3+) and a number of mono- and disubstituted chromium(III) ammine compounds. This study concerns a detailed analysis of the ligand field spectra of the complexes. The ab initio results are discussed within the framework of a ligand field analysis of the different compounds and are compared with the available experimental data. The basic ligand field assumptions (additivity and transferability postulates) are confirmed. The theoretical ligand field parameters reflect the same relative donor properties of the ligands as the semiempirical ones. Yet - as a partial amendment of the ligand field picture - the result of a population analysis reveals that the ligand to metal donations show rather significant variations from one complex to another. In comparison with experiment, the quartet-quartet transitions are reasonably well predicted. The spin-forbidden excitations are calculated to be too large, but the ab initio method agrees with experiment in identifying the trans difluoro compound as a 2T1 emitter and the other substituted complexes as 2E emitters. Moreover, at the ab inito level, insight is gained into the origin of the splitting of the 2E(g) parentage in tetragonal chromium(III) complexes.