With the aim to develop a rapid and nondestructive method to assess oxidation level of sunflower oil, fluorescence spectra were collected by a self-developed 365nm laser induced fluorescence (LIF) system. Both thermal oxidation at 60℃ and room-temperature degradation were implemented to obtain sunflower oil with different peroxide values (PV). Fluorescence peaks at about 440nm, 495nm, 520nm and 680nm were identified. Sunflower oil with lower PVs had lower fluorescence intensities. A control- part was designed in order to detect the fluorescence response at different spectral acquisition times. In contrast to the LIF spectra acquired from 10s to 60s (with a interval of 10s), the spectra acquired at 1s appeared to have the strongest response, and therefore it was used in the following part. After SG smoothing of the original LIF spectra, three different pre-processing methods were operated to correct the spectra. Top six normalized weights were extracted for evaluating the correlation between fluorophore in sunflower oil and PV. The peaks for hydroperoxides, chlorophyll and vitamin E had a very close association with the PV. The best partial least squares regression resulted in R2c of 0.9995 and R2p of 0.9972 and RMSEC of 0.0088g/(100g) and RMSEP of 0.0227g/(100g) for MSC pre-processing. The accuracy of the proposed LIF system was validated by using an independent set of samples with an R2 of 0.9681 and RMSE of 0.0411g/(100g). The absolute residual values below the national standard limit were all below 0.08g/(100g). These findings demonstrated that the proposed LIF system could be used to detect PV of sunflower oil with high accuracy, thus providing technical support for subsequent design of portable device and construction of edible oil oxidation warning system.