Protein control of cobalt-axial nitrogen ligand bond length has been proposed to modulate the reactivity of vitamin B(12) coenzyme during the catalytic cycle of B(12)-dependent enzymes. In particular, hyper-long Co-N bonds may favor homolytic cleavage of the trans-cobalt-carbon bond in the coenzyme. X-ray crystallographic studies point to hyper-long bonds in two B(12) holoenzymes; however, mixed redox and ligand states in the crystals thwart clear conclusions. Since EPR theory predicts an increase in Co(II) hyperfine splitting as donation from the axial N-donor ligand decreases, EPR spectroscopy could clarify the X-ray results. However, the theory is apparently undermined by the similar splitting reported for the 2-picoline (2-pic) and pyridine (py) adducts of Co(II) cobinamide (Co(II)Cbi(+)), adducts thought to have long and normal Co-N axial bond lengths, respectively. Cobinamides, with the B(12) 5,6-dimethylbenzimidazole loop removed, are excellent B(12) models. We studied Co(II)Cbi(+) adducts of unhindered 4-substituted pyridines (4-X-py's) in ethylene glycol to separate orbital size effects from Co-N axial distance effects on these splittings. The linear increase in splitting with the decrease in 4-X-py basicity found is consistent with the theoretically predicted increase in unpaired electron spin density as axial N lone pair donation to Co(II) decreases. No adduct (and hence no hyper-long Co(II)-N axial bond) was formed even by 8 M 2-pic, if the 2-pic was purified by a novel Co(III)-affinity distillation procedure designed to remove trace nitrogenous ligand impurities present in 2-pic distilled in the regular manner. Adducts formed by impurities in 2-pic and other hindered pyridines misled previous investigators into attributing results to adducts with long Co-N bonds. We find that many 2-substituted py's known to form adducts with simple synthetic Co models do not bind Co(II)Cbi(+). Thus, the equatorial corrin ring sterically impedes binding, making Co(II)Cbi(+) a highly selective binding agent for unhindered sp(2) N-donor ligands. Our results resolve the apparent conflict between EPR experiment and theory. The reported Co(II) hyperfine splitting of the enzyme-bound cofactor in five B(12) enzymes is similar to that of the relevant free cofactor. The most reasonable interpretation of this similarity is that the Co-N axial bond of the bound cofactor is not hyper-long in any of the five cases.