There were many deficiencies in the accuracy of soil resistivity test technology, and electrode arrangement is an important reason, especially in two-electrode method. If the contact between electrode and soil and the distance of two electrodes can be controlled, electrode influences will be avoided to the greatest extent. Therefore, based on two-electrode test model, zinc-contaminated sand was selected as simulated soil to test its AC resistivity. Electrode arrangement conditions of contact degree and electrode distance were set up, and the influences of them on resistivity measurement were studied combining with COMSOL simulation. The influences of electrode arrangement on resistivity test were studied, and an analytical model and error calculation formulas were established. The results show that test and simulation can verify and supplement each other. The greater contact degree, the more favorable resistivity test, and the influence of contact on resistivity test gradually decreases as it increases. The greater electrode distance, the weaker effect on resistivity test and the slender sample is more beneficial for improving the test accuracy. Conductive model is divided into two transition segments and normal segments in series based on whether it is affected by polarization. The length and polarization of each segment are characterized by radial current density, which decreases with the increase of contact, but slightly affected by electrode distance. Normal resistivity can be obtained by fitting the testing results under different electrode distances using the electrode arrangement model.