In order to provide references for heavy metals diagnosis of Vetiver grass, Raman spectroscopy technology was employed, the partial least squares (PLS) quantitative analysis model of heavy metal lead of Vetiver grass root was established, different processing methods were used to optimize the Raman spectra of Vetiver grass roots, Successive projection algorithm (SPA) was applied to screen the bands of Raman spectrum and enhance the model’s accuracy. The best Raman spectroscopy quantitative analysis model of lead content in Vetiver grass root were set up, 20 unknown samples were used to test the quality of optimized model. The prediction correlation coefficient and the root mean square error were 0.607 and 0.040 g/kg, respectively.