Quantum chemistry calculations of CASSCF/CASPT2 level together with ligand field analysis are used for the investigation of magnetic anisotropy of [Mo(CN)(7)](4-) complexes. We have considered three types of heptacyano environments: two ideal geometries, a pentagonal bipyramid and a capped trigonal prism, and the heptacyanomolybdate fragment of the cyano-bridged magnetic network K-2[Mn(H2O)(2)](3)[Mo(CN)(7)](2)(.)6H(2)O. At all geometries the first excited Kramers doublet is found remarkably close to the ground one due to a small orbital energy gap in the ligand field spectrum, which ranges between a maximal value in the capped trigonal prism (800 cm(-1)) and zero in the pentagonal bipyramid. The small value of this gap explains (i) the axial form of the g tensor and (ii) the strong magnetic anisotropy even in strongly distorted complexes. Comparison with available experimental data for the g tensor of the mononuclear precursors reveals good agreement with the present calculations for the capped trigonal prismatic complex and a significant discrepancy for the pentagonal bipyramidal one. The calculations for the heptacyanomolybdate fragment of K-2[Mn(H2O)(2)](3)[Mo(CN)(7)](2)(.)6H(2)O give g perpendicular to/g parallel to P approximate to 0.5 and the orientation of the local anisotropy axis close to the symmetry axis of an idealized pentagonal bipyramid. These findings are expected to be important for the understanding of the magnetism of anisotropic Mo(III)-Mn(II) cyano-bridged networks based on the [Mo(CN)(7)](4-) building block.