Fully buried portal frame-shaped piles have been widely used to control and stabilize large-scale landslides or slopes. However, the working mechanism of portal frame piles, such as the soil–pile interaction and load transfer behavior, is still not fully understood, which has limited the development of effective pile designs and construction. In this study, a well-designed laboratory model test was conducted in a sandy soil slope with a man-made slip surface. Nine portal frame piles were installed, synthetically considering the pile spacing between the rear and front piles as well as the support condition of the pile ends. To make the test closer to reality, in addition to a small thrust due to the excavation of the slope, greater landslide thrusts were imposed on the rear side of the portal frame piles by gradually applying static loads on the top of the slope. It is found that in addition to a concentrated thrust transferred via the cap beam, a landslide thrust accounting for 30 to 60 % of the landslide thrust on the rear pile was transmitted to the front pile by the soil sandwiched between the rear and front piles, which should not be ignored in designs. Meanwhile, to evenly mobilize the bearing capacity of both the rear and front piles, the reasonable spacing between them should be 1.0 to 1.5 times the diameter of the pile. The study also shows that portal frame piles should be embedded in a stable soil layer rather than the bedrock if they can meet the requirements for bearing capacity.