This paper established an optimal analytical model for rapid and simultaneous detection of sodium benzoate and potassium sorbate in cocktail by using surface-enhanced Raman technique, based on the self-built laboratory Raman point scanning system. Silver colloidal nanoparticles were prepared by reduction of silver nitrate with sodium citrate. In addition, the Raman spectra of sodium benzoate and potassium sorbate standard were compared with surface-enhanced Raman spectroscopy of cocktail samples containing sodium benzoate and potassium sorbate. 846.1, 1 007 and 1 605 cm-1 were surface-enhanced Raman characteristic shifts of sodium benzoate. 1 164, 1 389 and 1 651 cm-1 were surface-enhanced Raman characteristic shifts of potassium sorbate. At the same time, the stability of the peak in the Raman characteristic shift and the interrelation of Raman shift peak strength of two kinds of preservatives in the cocktail were analyzed. The detection method was of high stability, and the two kinds of preservatives were basically no interaction. Moreover, 42 sodium benzoate concentrations of 0.154 3~1.5 g·kg-1 and 45 potassium sorbate concentrations of 0.062~1.5 g·kg-1 were prepared to establish a linear regression model in cocktail. The best prediction models for sodium benzoate and potassium sorbate were two respective linear regression model established by 1 007, 1 605 and 1 164 1 651 cm-1. Besides, 43 different concentrations of sodium benzoate and potassium sorbate cocktail samples were predicted by the best prediction models. The result showed that the correlation coefficients (r) of sodium benzoate and potassium sorbate were 0.949 3, 0.921 8, root mean square error (RMSE) were respectively 0.088 2 and 0.142 9 g·kg-1, respectively. The rapid and quantitative determination of sodium benzoate and potassium sorbate in commercial cocktail can be realized by surface enhanced Raman spectroscopy. The results provides technical support for simultaneous and real-time online monitoring of preservatives in liquid food.