Deterioration of deep frying oils used by street vendors is one of the major food safety concerns. Fourier transform infrared (FTIR) spectroscopy coupled with partial least squares (PLS) regression was applied for rapid evaluation of the quality of deep frying oils collected from different street vendors (n = 109) using various frying processes in Shanghai. The levels of free fatty acids (FFA), total polar compounds (TPC), and viscosity of oils were determined with conventional methods and used as reference values for developing PLS models. The FFA (0.07–1.78 mg (KOH)/g) of all tested frying oils were below the maximum allowed value, while 5.5 % of oils had TPC (3.19–54.47 %) above the maximum allowed value (27 %) based upon the related chinese standards. FTIR coupled with PLS regression resulted in less satisfying results for FFA determination (predictability for soybean oils: R2 = 0.709, SEP = 0.14, RPD = 1.83), but showed great promise for rapid determination of viscosity (model predictability: R2 = 0.921–0.945, SEP = 0.68–0.71, RPD = 3.54–3.98) and TPC (predictability for soybean oils: R2 = 0.790–0.931, SEP = 1.89–2.94, RPD = 2.16–3.55) of frying oils from different commercial settings. Developing separated PLS models for shortening and non-shortening oils improved predictability, particularly for the analysis of TPC.