Heavy mental ionsin plants always have complex chelation with the organic molecular groups that have the near-infrared spectral (NIRS) absorptions. Therefore heavy mental ions in plants can be indirectly detected by using INRS technique basing on the chelation. An application of near infrared spectral technology fast detecting heavy metal lead (Pb) in vetiver grass leaves was analyzed. Combined with partial least squares (PLS), different preprocessing methods including smoothness, standard normal variate, baseline correction, multiplicative scatter correction, first derivative and second derivative were compared, model parameters were optimized by different wavelength selection methods including genetic algorithm, interval partial least square and successive projections algorithm, established the fast detection models of heavy metal Pb in vetiver grass leaves. The results showed that the external validation determination coefficient (R2) and root mean square error of prediction (RMSEP) was 0.87 and 0.18 separately. The study shows that the fast detection of heavy metal Pb in vetiver grass leaves using INRS technique is feasible.