This work presents calculations on small molecules using second-order Moller-Plesset perturbation theory with a generalized valence bond reference wave function. Moller-Plesset perturbation theory applied to a generalized valence bond reference (GVB-MP2) currently provides the best tradeoff between accuracy and computational feasibility among the methodologies of electronic structure. Frienser and co-workers have shown that the computational effort required for the GVB-MP2 methodology scales as no more than the third power of the size of the system, while that for the coupled-cluster and complete-active space methods scales as the seventh or worse power of system size. The GVB wave function is a qualitative wave function. Spectroscopic parameters and energetics at the GVB level are in qualitative agreement with experimentally determined values. The calculations presented in this work demonstrate that spectroscopic parameters computed using GVB-MP2 are in better agreement with experiment than those computed at the GVB level, and in close agreement with those obtained from the coupled-cluster plus singles and doubles with triples substitutions, which requires significantly more effort. (C) 2001 American Institute of Physics.