Despite the widespread use of stable isotopes in studies of avian movement, key assumptions on which the methodology is based remain unsubstantiated, including the assumption that measurements of stable hydrogen isotopes in feathers (δD_f) are consistent across time within the same laboratory or among laboratories using the same analytical protocols and keratin standards. We tested this assumption by remeasuring δD_f from 211 raptor feathers within and between laboratories. Initial and repeat samples were prepared and analyzed using identical protocols but analyzed in distinct automated runs with laboratory staff blind to sample identity. Reproducibility of δD_f measurements varied significantly and substantially among nine independent sample groups. Feather δD measurements among sample groups exhibited average isotopic shifts from -15.6%‰ to 27.5%‰ (an absolute difference of 43.1%‰), with standard deviations from 6.0%‰ to 12.4%‰. Therefore, despite existing analytical protocols to address issues of reproducibility, empirical data suggest that comparing δD_f measurements among studies or labs and pooling samples analyzed during different automated runs within a laboratory remain problematic. More importantly, poor reproducibility compromises the geographic assignment of origins based on δD_f , because the substantial differences in δD_f measurements between automated runs can result in spurious inferences regarding the origins of migratory birds. We caution against the continued use of δD_f for predicting geographic origin, and for addressing important conservation questions, until the factors affecting poor reproducibility are identified and improved reproducibility is demonstrated within and among laboratories across time and taxa.