Abstract: Unmanned aerial vehicle(UAV)remote sensing equipment is the first choice for disaster emergency monitoring because of its small, highly mobile and environmentally adaptable characteristics. In order to clarify the application prospects of UAV hyperspectral remote sensing in river oil spill emergency monitoring, six-rotor UAV, headwall imaging spectrometer (400-1000 nm) and Xinjiang crude oil were used as platform, detection instrument, and oil spill products respectively, and swimming ring with inner diameter of 45 cm was used as fence. A hyperspectral remote sensing experiment based on UAV for detecting oil spill information in rivers was carried out. The simulated oil film thickness ranged from 6.29 μm to 125.82 μm, obtaining headwall hyperspectral data of 30 groups of oil film with different thickness and a scene of oil spill channel hyperspectral data. By analyzing the correlation between oil film thickness and headwall reflectance spectral band, 984 nm was determined as the optimum spectral response band of oil film thickness. A model for estimating oil film thickness on water surface based on 984 nm spectral reflectance was established to verify the maximum relative error between predicted oil volume and actual oil volume, which was 7.8%. It was considered that it was feasible to detect oil spill in rivers by using hyperspectral sensors on UAV. For the very thin oil film formed in the later stage of the experiment(about 1 μm), which exceeded the lower limit of model detection, the oil film thickness was estimated by referring to Bonn agreement, and the oil spill was calculated. It was found that the actual oil used was in the predicted oil volume range, meaning it was feasible to estimate the oil spill in the very thin oil film area of the river by using Bonn agreement.