The structural proteins that comprise ∼90% of animal hair have the potential to record environmentally and physiologically determined variation in δ2H and δ18O values of body water. Broad, systematic, geospatial variation in stable hydrogen and oxygen isotopes of environmental water and the capacity for rapid, precise measurement via methods such as high-temperature conversion elemental analyzer/isotope ratio mass spectrometry (TC/EA-IRMS) make these isotope systems particularly well suited for applications requiring the geolocation of hair samples. In order for such applications to be successful, however, methods must exist for the accurate determination of hair δ2H and δ18O values reflecting the primary products of biosynthesis. Here, we present the results of experiments designed to examine two potential inaccuracies affecting δ2H and δ18O measurements of hair: the contribution of non-biologic hydrogen and oxygen to samples in the form of sorbed molecular water, and the exchange of hydroxyl-bound hydrogen between hair keratin and ambient water vapor. We show that rapid sorption of molecular water from the atmosphere can have a substantial effect on measured δ2H and δ18O values of hair (comprising ∼7.7% of the measured isotopic signal for H and up to ∼10.6% for O), but that this contribution can be effectively removed through vacuum-drying of samples for 6 days. Hydrogen exchange between hair keratin and ambient vapor is also rapid (reaching equilibrium within 3–4 days), with 9–16% of the total hydrogen available for exchange at room temperature. Based on the results of these experiments, we outline a recommended sample treatment procedure for routine measurement of δ2H and δ18O in mammal hair. Copyright © 2005 John Wiley & Sons, Ltd.