The full reaction path for the conversion of carbon dioxide to hydrogencarbonate has been computed at the B3LYP/6-311+G** level, employing a [(NH(3))(3)Zn(OH)](+) model catalyst to mimic the active center of the enzyme. We paid special attention to the question of how the catalytic cycle might be closed by retrieval of the catalyst. The nucleophilic attack of the catalyst on CO(2) has a barrier of 5.7 kcal mol(-1) with inclusion of thermodynamic corrections and solvent effects and is probably the rate-determining step. This barrier corresponds well with prior experiments. The intermediate result is a Lindskog-type structure that prefers to stabilize itself via a rotation-like transition state to give a Lipscomb-type product, which is a monodentate hydrogencarbonate complex. By addition of a water molecule, a pentacoordinated adduct with pseudo-trigonal-bipyramidal geometry is formed. The water molecule occupies an equatorial position, whereas the hydrogencarbonate ion is axial. In this complex, proton transfer from the Zn-bound water molecule to the hydrogencarbonate ion is extremely facile (barrier 0.8 kcal mol(-1)), and yields the trans,trans-conformer of carbonic acid rather than hydrogencarbonate as the leaving group. The carbonic acid molecule is bound by a short O...H-O hydrogen bond to the catalyst [(NH(3))(3)Zn(OH)](+), in which the OH group is already replaced by that of an entering water molecule. After deprotonation of the carbonic acid through a proton relay to histidine 64, modeled here by ammonia, hydrogencarbonate might undergo an ion pair return to the catalyst prior to its final dissociation from the complex into the surrounding medium.