Taking the serum obtained from daily clinical inspection as study object, the Raman spectra were collected with the range of 250 to 2 400 cm-1. Taking the serum cholesterol content determined by traditional clinical biochemical method as contrast, together with combination of PLS, we established a Raman spectroscopy quantitative models for cholesterol content in serum. The experimental results show that Raman spectroscopy displacement could reflect certain chemical group in organic compound. The correction model of cholesterol content in serum was established with the coefficient R=0.909 6, cross-validation calibration standard deviation (RMSECV) of 0.24, standard error of prediction of test set (RMSEP) was 0.69, and coefficient (R) is 0.926 2. The results showed that application of Raman spectroscopy combined with chemometrics in the rapid noninvasive detection of serum cholesterol content is feasible.