A low-confinement mode discharge which optimizes the capability of the new main-ion charge-exchange-recombination spectroscopy system on DIII-D (Luxon 2002 Nucl. Fusion42 614) to measure deuterium toroidal velocity is interpretted in comparison with the predictions of neoclassical theory, with an emphasis on the plasma edge region. The observed peaking in the deuterium toroidal velocity near the separatrix is shown to be consistent with intrinsic co-rotation due to ion orbit loss. In general, the standard neoclassical toroidal and poloidal momentum transport rates are found to be smaller than those inferred from experiment.