Based on time-resolved fluorescence of oils, an oceanographic fluorescence Lidar was designed to identify oil pollutions. A third harmonic (at 355nm) of Nd:YAG laser is used as the excitation source, and the fluorescence intensities and lifetimes of oil fluorescence at wavelength from 380 nm to 580 nm are measured by an intensified CCD (ICCD). In the experiments, time-resolved fluorescence spectra of 20 oil samples, including crude oils, fuel oils, lubricating oil, diesel oils and gasoline, are analyzed to discuss fluorescence spectral characteristics of samples for oil classification. The spectral characteristics of oil fluorescence obtained by ICCD with delay time of 2 ns, 4 ns, and 6 ns were studied by using the principal component analysis (PCA) method. Moreover, an efficient method is used to improve the recognition rate of the oil spill types, through enlarging spectral differences of oil fluorescence at different delay times. Experimental analysis shows that the optimization method can discriminate between crude oil and fuel oil, and a more accurate classification of oils is obtained by time-resolved fluorescence spectroscopy. As the result, comparing to traditional fluorescence spectroscopy, a higher recognition rate of oil spill types is achieved by time-resolved fluorescence spectroscopy which is also a feasibility technology for Ocean Lidar.