The hyperspectral imaging technology combined with BP neural network was used to detect pesticide residues on surface of fruit, crystal crown pears, such as chlorpyrifos and propargite pesticide. Twenty samples were prepared for both chlorpyrifos and propargite with different concentrations. Dropping the 100μL and 150μL pesticide solution droplets in pear surface, the hyperspectral images were scanned in 8354678~16483568nm band range. Twenty regions of interest (ROI) were available for four sample groups, from which five samples were randomly selected as predictor set. The results of models built by BP neural network for 100μL and 150μL pesticide solution region showed that correlation coefficient of the samples was greater than 099 and 095 respectively. RMSEC and RMSEP maximum value was respectively 06349, 13239 and 17425, 34417 The results indicated that 150μL pesticide solution region was better than 100μL in the modeling. Hyperspectral image technology combined with BP neural network method is feasible for detection of pesticide residues on pear surface . which provide a new method for detection of pesticide residues on fruit surface.