Abstract: This study aimed at validating the relationship between pesticide residue(PR) and hyperspectral absorption parameters, and used the selected hyperspectral absorption parameters to build an effective model to predict the chlorpyrifos residue levels in vegetatble samples. Firstly, reflectance spectral data for different chlorpyrifos pesticide concentration on Chinese chive samples were collected, then the correlation between reflectance's first-order differential value and the amount of pesticide residues (measured by GC-MS) were calculated, where 33 different kind hyperspectral absorption parameters were calculated and the correlation coefficients with the pesticide residue were also computed. The results revealed that the directly spectral reflectance and first-order differential value's correlation coefficients were low, except for near infrared bands between 789-867 nm and 1 860 nm. Among the 33 hyperspectral absorption parameters, the sum of first-order differential near infrared (SDnir) was chosen as a key factor. Finally, the pesticide concentration fits with first-order differential value's 1 860 nm (FD1860) and SDnir by linear, polynomial and exponential function. Then cross validation experiments were carried out to verify the reliability of the models. The final experiments showed that the SDnir quadratic polynomial model got the highest coefficient of determination R2 and lower RMSE. It is suggested that the hyperspectral absorption parameters could be a good indicator for pesticide residues on vegetable samples. The changes of SDnir directly reflect to the changes of pesticide residues. The experimental results show that it is feasible to use vegetable hyperspectral absorption parameters for predicting pesticide residues.