<p>Although oil film thickness is an important indicator for estimating the volume of oil spill pollution on the sea surface, the quantitative inversion on oil film thickness is still a challenge. This study proposes a quantitative inversion method for oil film thickness based on the three-band fluorescence index (TBFI). The fluorescence of oil film samples with different thicknesses was excited using a tripled-frequency Nd: YAG laser. The fluorescence spectra of the oil film samples with different thicknesses were measured by a high-resolution spectrometer, and the corresponding optimal band combinations were explored using the optimal band combination algorithm. The selected optimal band combinations were combined with partial least squares regression (PLSR) to establish an oil film thickness prediction model. By comparing the prediction results of four TBFI-PLSR models, TBFI3-PLSR showed the best prediction effect for oil film thickness. For gasoline and diesel in the TBFI3-PLSR model, R², RMSE, and MSRE are 0.989, 25.94, 0.80%; 0.908, 78.71, 67.45% respectively.</p>