To achieve the detection of multi-component pesticide residues, the Partial Least Square(PLS) method was applied to establishing a calibration model of fluorescence spectral measurement systems, and to predict the pesticide residues of acetamiprid by separating the overlapped spectrum in fluorescence spectroscopy of pesticide residues.By taking the predicted residual error square sum as an evaluating criterion, twenty characteristic wavelengths were selected and then the optimal number of principal components and the optimal analysis model were determined by the cross verification method. According to the test of prediction sets (100, 220, 450 mg/kg), predictive values of acetamiprid residues are 101.45, 222.91 and 440.08 mg/kg on the surface of filter paper, and 98.67, 208.56 and 419.22 mg/kg on the surface of tomato. The correlation coefficients between predictive values and true values respectively reach 0.996 and 0.988. The results demonstrate that the method using PLS in fluorescence spectral analysis for measuring the acetamiprid residue has good performance in shorter measuring time, nondestructive testing and higher accuracy and can effectively implement quantitive analysis for complex and multi-component systems.