Drag anchors have the high anchor efficiency and low manufacturing cost, and are widely used for the mooring of offshore floating platforms. However, there are often non-negligible errors in the prediction of the anchor penetration and holding capacity when the drag anchor is installed in sand. A modified method was developed to predict the installation process of the drag anchor, which divides the whole installation process into several incremental steps and allows the bottom surface of the fluke to separate from sand. At each incremental step, the limit equilibrium equations of the anchor and soil in failure are established to obtain the holding capacity, and the increment of the horizontal displacement, burial depth and rotation were calculated, based on the principle of minimum force to drag the anchor, with a reasonable direction specified for the drag anchor motion. Small-scale model tests were conducted on drag anchors with two types of the shank length, and the relative density of the sand sample was 47%. The holding capacity and position of the anchor were measured in real time. During the tests, it was observed that the bottom surface of the fluke already separated from the sand sample in the early stage of installation, and the burial depth of the anchor still increased with the increment of the holding capacity, which proved the rationality of the assumption of separation in the modified method. The rationality of the modified method was verified by comparing the measured values and predicted values of the anchor trajectory and holding capacity. Both the test results and calculation results indicate that, when the shank length increases from 30 cm to 36 cm, although the penetration depth of the anchor padeye and the rotation of the anchor body decrease, the final holding capacity still increases by 18%.