A semiempirical theoretical study of the [2+2] cycloaddition between ketene and formaldehyde AIP Conf.State-of-the-art ab initio quantum chemical techniques have been employed to ascertain the reaction path and associated energetics for the dissociation of CH 2 CO into 1 CH 2 ϩCO and thereby to investigate the kinetics of this dissociation via variational Rice-Ramsperger-Kassel-Marcus ͑RRKM͒ theory. The quantum chemical computations focused on the determination of geometric structures, energies, and force fields for four constrained C-C distances ͑2.2, 2.5, 2.8, and 3.1 Å͒ spanning the inner transition-state region. Optimized structures were obtained with the coupled-cluster singles and doubles method including a perturbative triples term ͓CCSD͑T͔͒, as implemented with a contracted ͓C/O, H͔ basis set of ͓5s4 p2d1 f , 4s2p1d͔ quality. The resulting energetics were corrected for basis set incompleteness and higher-order electron correlation with the aid of second-order Mo "ller-Plesset perturbation theory ͑MP2͒ predictions given by an immense ͓13s8 p6d4 f , 8s6p4d͔ basis combined with 6-31G* Brueckner doubles results augmented with perturbative contributions from both connected triple and quadruple excitations. Quadratic force fields along the reaction path were determined at the CCSD/͓5s4 p2d, 4s2p͔ level of theory. Anharmonic effects in the enumeration of accessible states for the transition state were accounted for by a direct statistics approach involving repeated MP2/6-31G* energy evaluations. Two separate reaction coordinates defined by the C-C bond length or alternatively the center-of-mass separation between the 1 CH 2 and CO fragments were explicitly considered in these direct statistical analyses. A spectroscopic quality quartic force field for ketene derived in a companion ab initio study was employed in the evaluation of the anharmonic reactant density of states. The final statistical predictions for the energy dependence of the dissociation rate constant are found to be in quantitative agreement with experiment ͑i.e., generally within 30%͒, thereby providing strong evidence for the quantitative validity of variational RRKM theory.