The excitation energies calculated by the HAM/3 procedure for ΠΠ* transitions in linear molecules can be internally inconsistent by as much as ± 0.6 eV. In the recent study by Åsbrink etal., the problem was avoided by adopting Recknagel's expressions and requiring the proper average ΠΠ* excitation energy. In this paper, we trace the small inconsistency back to its origin in HAM/3 theory and derive the analytical expression for the energy correction as well as Recknagel's formulas. Numerical examples studied include all seven linear molecules investigated by Åsbrink etal. The explicit expression for the correction enables us to perform meaningful configuration-interaction calculations on the excited states, as illustrated by the carbon suboxide molecule.