Ab initio, semiempirical, and HMO perturbation calculations were employed to assess the relative positioning of the closely situated low-lying unoccupied pi* MOs a(2)(chi) and b(1)(psi) in 1,10-phenanthroline (phen) and its 3,4,7,8-tetramethyl (tmphen) and four symmetrical diaza derivatives (n,m-dap). Compared to a(2)(chi), the b(1)(psi) pi MO is distinguished by markedly higher MO coefficients at the chelating nitrogen pi centers in 1,10-positions; eventually, a higher Coulomb integral value at those positions will thus always favor the lowering of b(1) beyond a(2). Using the Coulomb integral parameter at the chelating 1,10-nitrogen pi centers as the HMO perturbation variable, the crossing of both energy levels in terms of decreasing preference for the a(2)(chi) over the b(1)(psi) orbital as the lowest unoccupied MO follows the sequence 5,6-dap > 2,9-dap > 4,7-dap, phen > 3,8-dap. The calculations reveal a(2)(chi) as the LUMO in 5,6-dap for all reasonable perturbation parameters, in agreement with previous observations for ruthenium(II) complexes which reveal a discrepancy between the lowest-lying "redox pi* orbital" (a(2)) and the "optical pi* MO" (b(1)) to which the most intense low-energy MLCT transition occurs. According to the HMO calculations, the situation is more ambiguous for the 4,7-dap analogue, yet EPR/ENDOR studies clearly show that the one-electron-reduced ligand and its tetracarbonylmetal(0) complexes (Cr, Mo, W) have the b(1)(psi) orbital singly occupied. Only ab initio calculations with double-zeta basis and inclusion of d polarization functions reproduced correctly the experimentally observed orbital ordering for tmphen (a(2) < b(1)) and for phen and 4,7-dap (b(1) < a(2)).