Measurements of precipitation isotopic composition have been conducted on a daily basis for 1 yr at Bomi, in the southeast Tibetan Plateau, an area affected by the interaction of the southwest monsoon, the westerlies, and Tibetan high pressure systems, as well as at Lhasa, situated west of Bomi. The measured isotope signals are analyzed both on an event basis and on a seasonal scale using available meteorological information and airmass trajectories. The processes driving daily and seasonal isotopic variability are investigated using multidecadal climate simulations forced by twentieth-century boundary conditions and conducted with two different isotopic atmospheric general circulation models [the isotopic version of the Laboratoire de Météorologie Dynamique GCM (LMDZiso) and the ECHAM4iso model]. Both models use specific nudging techniques to mimic observed atmospheric circulation fields. The models simulate a wet and cold bias on the Tibetan Plateau together with a dry bias in its southern part. A zoomed LMDZ simulation conducted with ~50-km local spatial resolution dramatically improves the simulation of isotopic compositions of precipitation on the Tibetan Plateau. Simulated water isotope fields are compared with new data and with previous observations, and regional differences in moisture origins are analyzed using back-trajectories. Here, the focus is on relationships between the water isotopes and climate variables on an event and seasonal scale and in terms of spatial and altitudinal isotopic gradients. Enhancing the spatial resolution is crucial for improving the simulation of the precipitation isotopic composition.