OBJECTIVE： To provide theoretical basis for the detection of paracetamol by infrared spectroscopy. METHODS： 0.002-0.02 g paracetamol was accurately weighed， added into 0.2 g potassium bromide， respectively， fully ground and well mixed， potassium bromide pressing plate method was used for to-be-tested test sample. Using hollow cathode lamp as light， scanning frequency was 30 times， resolution was 0.5 cm－1， scan range was 4000-400 cm－1， infrared absorption spectrum was determined and recorded to select the characteristic absorption peaks， then determine the optimized characteristic absorption peaks. Linear and nonlinear models were respectively established by using mathematics modeling methods. RESULTS：1 016 cm－1 absorption peak with weaker absorbance but stronger features was selected for the analysis. According to the model establishing and calculation， the accuracy of the nonlinear model was much higher than the linear model，r＝0.942. CONCLUSIONS： Nonlinear quantitative model for quantitatively determining the content of paracetamol is feasible， and suitable for the on-destructive and rapid on-line quality control of paracetamol.