We investigate basis set convergence for a series of density functional theory (DFT) functionals (both hybrid and nonhybrid) and compare to coupled-cluster with single and double excitations and perturbative triples [CCSD(T)] benchmark calculations. The case studied is the energetics of the water oxidation reaction by an iridium-oxo complex. Complexation energies for the reactants and products complexes as well as the transition state (TS) energy are considered. Contrary to the expectation of relatively weak basis set dependence for DFT, the basis set effects are large, for example, more than 10 kcal mol−1 difference from converged basis for the activation energy with “small” basis sets (DZ/6-31G** for Ir/other atoms, or SVP) and still more than 6 kcal mol−1 for def2-TZVPP/6-31G**. Inclusion of the dispersion correction in DFT-D3 schemes affects the energies of reactant complex (RC), TS, and product complex (PC) by almost the same amount; it significantly improves the complexation energy (the formation of RC), but has little effect on the activation energy with respect to RC. With converged basis, some pure GGAs (PBE-D3, BP86-D3) as well as the hybrid functional B3LYP-D3 are very accurate compared to benchmark CCSD(T) calculations. © 2012 Wiley Periodicals, Inc.