We find that regions of the excited state potential energy surface of formaldimine, which are accessible from the Franck-Condon configuration, are incorrectly described by the restricted open-shell Kohn-Sham (ROKS) approach. In these regions, the deviations of the ROKS energies from the time-dependent density functional theory results are not a simple shift. Contrary to what is argued in the Comment by Doltsinis and Fink [J. Chem. Phys.XX, XXX (2004)], these differences can play a role in the excited state molecular dynamics of formaldimine at finite temperature.