Activities of POD in tomato leaves were measured rapidly using hyperspectral imaging technology combined with chemometrics method. Operation process was: extracting spectra curve, pretreatment of spectra data, extracting characteristic wavelengths with SPA, and establishing prediction model for determining POD activities. In comparison with other methods such as SG, SNV, MSC, 1-Der and 2-Der, DOSC was the optimal pretreatment. It was shown in this research that SPA-PLS model was the optimal effective model among all models (SPA-MLR, SPA-PLS, SPA-BPNN and SPA-LS-SVM) for forecasting POD activities. The model was based on reflectance information of effective wavelengths (443, 464, 413, 410, 401, 402, 426 and 926 nm) extracted by SPA. Rp and RMSEP were 0.9353 and 37.80 U x g(-1), respectively. The result indicated that it was feasible to determine the POD activities with hyperspectral imaging technology, and the prediction accuracy of model was satisfactory. It was a new method for dynamic observation of POD activities and growth state of tomato.