The feasibility of the portable X-ray fluorescence (PXRF) spectrometer in ex-situ measuring soil heavy metals in the laboratory was determined by comparison of the traditional ICP-AES method. 70% of soil samples were used to establish the linear calibration models of soil Cr, Cu, As and Pb concentrations and the remaining 30% of soil samples were used to validate the established calibration models. The results showed that the detection limits of PXRF were 17.7 mg/kg for Cr, 10.4 mg/kg for Cu, 5.4 mg/kg for As, and 6.2 mg/kg for Pb, respectively, all are below the national 1st standards for soil environmental quality. The reference materials from China and abroad were measured iteratively, and the residual predictive deviation ranged from –8.9% to 7.9%, which indicated PXRF has better precision and accuracy in measuring soil heavy metal content. The R² of the fitted curves between PXRF and ICP methods were 0.817 3, 0.787 0, 0.673 3 and 0.722 1, respectively, all reached the significant level, which suggested PXRF can ex-situ rapidly measure soil Cr, Cu, As and Pb contents in laboratory. The R² values of calibration models established by resampling were 0.912 4 for Cr, 0.897 9 for Cu, 0.772 3 for As, and 0.872 9 for Pb, respectively, which were gradually close to the ideal models and fully validated the feasibility of PXRF and the accuracy of the calibration curves. In general, PXRF can ex-situ rapidly determining soil Cr, Cu, As an Pb as well as contamination screening. However, a calibration curve must be established when using PXRF in measuring soil As in the laboratory because of the lower of measure accuracy.