Devices based on deep-blue emitting iridium(III) complexes with N-heterocyclic carbene (NHC) ligands have recently been shown to give excellent performance as phosphorescent organic light-emitting diodes (PHOLEDs). To facilitate the design of even better deep-blue phosphorescent emitters, we carried out density functional theory (DFT) calculations of the lowest triplet (T-1) potential-energy surfaces upon lengthening the iridium-ligand (Ir-C) bonds. Relativistic time dependent DFT calculations demonstrate that this changes the nature of T-1 from a highly emissive metal-to-ligand charge transfer ((MLCT)-M-3) state to a metal centered ((MC)-M-3) state where the radiative decay rate is orders of magnitude slower than that of the (MLCT)-M-3 state. We identify the elongation of an Ir-C bond on the NHC group as the pathway with the lowest energy barrier between the (MLCT)-M-3 and (MC)-M-3 states for all complexes studied and show that the barrier height is correlated with the experimentally measured nonradiative decay rate. This suggests that the thermal population of (MC)-M-3 states is the dominant nonradiative decay mechanism at room temperature. We show that the (MLCT)-M-3 -> (MC)-M-3 transition is reversible, in marked contrast to deep-blue phosphors containing coordinating nitrogen atoms, where the population of (MC)-M-3 states breaks Ir-N bonds. This suggests that, as well as improved efficiency, blue PHOLEDs containing phosphors where the metal is only coordinated by carbon atoms will have improved device lifetimes.