In response to the Deepwater Horizon oil spill in the Gulf of Mexico, a method of near-infrared imaging spectroscopic analysis was developed to map the locations of thick oil floating on water. Specifically, this method can be used to derive, in each image pixel, the oil-to-water ratio in oil emulsions, the sub-pixel areal fraction, and its thicknesses and volume within the limits of light penetration into the oil (up to a few millimeters). The method uses the shape of near-infrared (NIR) absorption features and the variations in the spectral continuum due to organic compounds found in oil to identify different oil chemistries, including its weathering state and thickness. The method is insensitive to complicating conditions such as moderate aerosol scattering and reflectance level changes from other conditions, including moderate sun glint. Data for this analysis were collected by the NASA Airborne Visual Infrared Imaging Spectrometer (AVIRIS) instrument, which was flown over the oil spill on May 17, 2010. Because of the large extent of the spill, AVIRIS flight lines could cover only a portion of the spill on this relatively calm, nearly cloud-free day. Derived lower limits for oil volumes within the top few millimeters of the ocean surface directly probed with the near-infrared light detected in the AVIRIS scenes were 19,000 (conservative assumptions) to 34,000 (aggressive assumptions) barrels of oil. AVIRIS covered about 30 percent of the core spill area, which consisted of emulsion plumes and oil sheens. Areas of oil sheen but lacking oil emulsion plumes outside of the core spill were not evaluated for oil volume in this study. If the core spill areas not covered by flight lines contained similar amounts of oil and oil-water emulsions, then extrapolation to the entire core spill area defined by a MODIS (Terra) image collected on the same day indicates a minimum of 66,000 to 120,000 barrels of oil was floating on the surface. These estimates are preliminary and subject to revision pending further analysis. Based on laboratory measurements, near-infrared (NIR) photons penetrate only a few millimeters into oil-water emulsions. As such, the oil volumes derived with this method are lower limits. Further, the detection is only of thick surface oil and does not include sheens, underwater oil, or oil that had already washed onto beaches and wetlands, oil that had been burned or evaporated as of May 17. Because NIR light penetration within emulsions is limited, and having made field observations that oil emulsions sometimes exceeded 20 millimeters in thickness, we estimate that the volume of oil, including oil thicker than can be probed in the AVIRIS imagery, is possibly as high as 150,000 barrels in the AVIRIS scenes. When this value is projected to the entire spill, it gives a volume of about 500,000 barrels for thick oil remaining on the sea surface as of May 17. AVIRIS data cannot be used to confirm this higher volume, and additional field work including more in-situ measurements of oil thickness would be required to confirm this higher oil volume. Both the directly detected minimum range of oil volume, and the higher possible volume projection for oil thicker than can be probed with NIR spectroscopy imply a significantly higher total volume of oil relative to that implied by the early NOAA (National Oceanic and Atmospheric Administration) estimate of 5,000 barrels per day reported on their Web site.