Abstract: Abstract: Determining the pesticide residues in fruits is of great significance to identify the edible safety of food for better sales volume. However, the non-destructive technology is still lacking for the detection of pesticide residues in apples, particularly for high efficiency, cost-saving, and easy operation. In this study, a new non-destructive analysis was developed to detect the pesticide residues using Raman spectroscopy. It also highly contributed to the rapid non-destructive detection and quantitative evaluation of pesticide residues on fruits. Surface Enhanced Raman Spectroscopy (SERS) was utilized to explore, where a new nicotinic pesticide with acetamiprid for pest control in apple production was taken as the research object, whereas, the rapid detection of apple pesticide residues using silver sol (AgNPs) as the enhanced substrate. The polymer sodium polyacrylate was added to the silver sol as a stabilizer, further to prevent the oxidation and deposition of silver sol during long-time storage. The pH value of silver sol was adjusted to change the adsorption state of pesticide molecules on the surface of silver colloidal particles. The silver sol presented the best enhancing performance on the Raman scattering of acetamiprid pesticidewhen the pH value was 6.5. Furthermore, 1 mol/L NaCl solution as a coagulant and mixed with silver sol in a ratio of 1:5 was greatly improved the SERS effect of acetamiprid, further promoting the adsorption between pesticide and silver colloidal particles. The SERS spectra of acetamiprid collected by the improved silver sol showed that there was a great enhancement in the effect of improved silver sol on Raman scattering. Specifically, the coefficient of variation of the SERS spectrum was reduced from 0.0625 to 0.0307, and the relative standard deviations of characteristic peak intensity at 627, 835, and 1107 cm-1 were 6.14%, 6.83%, and 6.99%, respectively. The minimum detection limit of acetamiprid was reduced from 0.683 to 0.035 mg·kg-1. The improved silver sol was used to collect the SERS spectrum of apple samples containing acetamiprid. Different pretreatment and modeling were used to establish the prediction model of acetamiprid residue concentration on the surface of apple samples. The results showed that: A prediction model of acetamiprid pesticide residue in the apple was successfully established using the Kalman smoothing (Rauch-Tung-Striebel, RTS), fluorescence background deduction (asymmetrically reweighted Penalized Least Squares, arPLS), and extended multiplicative scattering correction (Extended Multiplicative Signal Correction, EMSC) pretreatment combined with partial least squares in the Raman spectral range of 400-2300 cm-1. The predicted correlation coefficient (prediction coefficient, Rp) was 0.974, the root mean square error of prediction (Root Mean Square Error of predictions, RMSEp) was 0.0441 mg/kg, the corrected correlation coefficient (correlation coefficient, Rc) was 0.986, and the corrected root mean square error (Root Mean Square Errors of calibration, RMSEc) was 0.0369 mg/kg. When collecting signals from apples, in which the content of acetamiprid was in the range of 0.012 mg/kg and 10.830 mg/kg, the lowest detection limit of acetamiprid in the apple was 0.035 mg/kg, one order of magnitude lower than the detection limit of 0.8 mg/kg in the national standard. Consequently, the SERS technology can widely be expected to qualitatively and quantitatively analyze acetamiprid pesticide residues in apples.