The lateral resistance of pile foundations has typically been based on the resistance of circular pipe piles. In addition, most instrumented lateral load tests and cases history have involved circular piles. However, piles used in engineering practice may also be non-circular cross-section piles such as square and H piles. Some researchers have theorized that the lateral resistance of square piles will be higher than that of circular piles (Reese and Van Impe, 2001; Briaud et al, 1983; Smith, 1987) for various reasons, but there is not test data to support this claims. To provide basic comparative performance data, lateral load tests were performed on piles with circular, square and H sections. To facilitate comparisons, all the tests piles were approximately 12 inches in width or diameter and were made of steel. The square and circular pipe sections had comparable moments of inertia; however, the H pile was loaded about the weak axis, as is often the case of piles supporting integral abutments, and had a much lower moment of inertia. The granular fill around the pile was compacted to approximately 95% of the standard Proctor maximum density and would be typical of fill for a bridge abutment. Lateral load was applied with a free-head condition at a height of 1 ft above the ground surface. To define the load-deflection response, load was applied incrementally to produce deflection increments of about 0.25 inches up to a maximum deflection of about 3 inches. Although the square and pipe pile sections had nearly the same moment of inertia, the square pile provided lateral resistance that was 20 to 30% higher for a given deflection. The lateral resistance of the H pile was smaller than the other two pile shapes but higher than what it is expected based on the moment of inertia. Back analysis with the computer program LPILE indicates that the pile shape was influencing the lateral resistance. Increasing the effective width to account for the shape effect as suggested by Reese and Van Impe (2001) was insufficient to account for the increased resistance. To provide agreement with the measured response, p-multipliers of 1.2 and 1.35 were required for the square pile and H piles, respectively. The analyses suggest that the increased resistance for the square and H pile sections was a result of increases in both the side shear and normal stress components of resistance. Using the back-calculated p-multipliers provided very good agreement between the measured and computed load-deflection curves and the bending moment versus depth curves.