For the rapid detection of carbaryl residue in duck meat, synchronous fluorescence spectroscopy was used, and GA combined with SVR was used to establish regression forecasting mode for the application of forecasting carbaryl residue in duck meat . Firstly, fluorescence spectrophotometer was used to get the 3D synchronous fluorescence spectra of carbaryl hydrolysate and duck solution containing carbaryl, and 140 nm was selected as the optimum wavelength difference Δλ; Secondly, some concentration quenching was analysed. Finally, GA was used to optimize and choose the 3D synchronous fluorescence spectra. According to the root mean square error of cross-validation (RMSECV) 21 characteristic wavelengths were chosen, then the full wavelength and 21 characteristic wavelengths were used as input characteristic variables of SVR regression forecasting model respectively. At last the results showed that characteristic wavelengths chosen by GA can get better forecasting results, and the correlation coefficient of the prediction samples set and the root mean squared error (RMSEP) were 0.976 4 and 12.232 2, respectively. The results of experiments showed that the synchronous fluorescence spectroscopy could be used to detect carbaryl residue in duck meat efficiently and rapidly when combined with GA-SVR.