Estimates of atmospheric moisture are critical for understanding the links and feedbacks between atmospheric CO and global climate. At present, there are few quantitative moisture proxies that are applicable to deep time. We present a new proxy for atmospheric moisture derived from modern climate and leaf biomarker data from North and Central America. Plants have a direct genetic pathway to regulate the production of lipids in response to osmotic stress, which is manifested in a change in the distribution of simple aliphatic lipids such as n-alkanes. The Average Chain Length (ACL) of these lipids is therefore statistically related to mean annual vapor pressure deficit (VPD), enabling quantitative reconstruction of VPD from sedimentary n-alkanes. We apply this transfer function to the Armantes section of the Calatayud-Daroca Basin in Central Spain, that spans the Middle Miocene Climatic Optimum (MMCO) and the Middle Miocene Climate Transition (MMCT). Reconstructed VPD rises from 0.13 to 0.92 kPa between 16.5 and 12.4 Ma, indicating a substantial drying through the MMCT. These data are consistent with fossil assemblages and mammalian stable isotope data, highlighting the utility of this new organic molecular tool for quantifying hydrologic variability over geologic timescales.