||In order to cater for the economic development of the outlying areas in Hong Kong, the KCRC West Rail project was proposed. This major infrastructure project requires deep foundation along the route, and a more economic and rational design approach was to be justified by conducting full-scale pile load tests. The lateral load tests on three single piles and three pile groups provided a precious opportunity to investigate the analysis of laterally loaded large-diameter bored piles. The design of a laterally loaded pile is usually carried out using elastic continuum method (Poulos 1980) and p-y method (Reese et al. 1974). These two common design methods have their own pros and cons, however, they share the common difficulty of selecting proper soil parameters for reliable prediction. In this research, the test results were thoroughly interpreted using a newly proposed simple technique. They were then compared with predictions using the common design methods, in order to study the applicability of the common design methods on the large-diameter bored piles and the suitable choice of parameters for soils in Hong Kong. The behaviour of laterally loaded test piles was examined in more details using numerical analysis. Firstly, a laterally loaded single pile was simulated using a 3D finite difference program (FLAC3D) to investigate the effect of the construction sequence and the kentledge loading on the soil and pile behaviour. Secondly, a parametric analysis of a pile group was conducted using a 3D analytical program (FLPIER) to investigate the effect of the axial soil response and the pile cap effectiveness on the group pile behaviour. In routine design, concrete cracking is usually considered using constant cracked pile stiffness along the whole pile length. This might resulting in overpredicted deflections and underestimated maximum bending moments. Hence, a simple technique, which incorporates the nonlinear concrete behaviour, is proposed to interpret the measured rotations along the pile lengths. The technique involves iterative procedures to make a best-fit to the measured rotation by using a fourth-order polynomial to represent the soil reaction profile and then carrying out integration. This technique can be used to back-analyse the values of angle of shearing resistance (φ') and constant of horizontal subgrade reaction (nh) for the p-y curves suggested by Reese et al. (1974). It is found that the two parameters vary linearly with the SPT 'N' values in logarithmic scale (i.e. φ' = 10logN+27; nh = 40logN-28). In addition, the comparisons between the measured test results and the predictions using the cornmon design methods for the single piles show that, both design methods give reasonable prediction for soils in Hong Kong provided that an appropriate choice of parameters is used. For the analysis of the pile groups, it is found that the deduced nh values, based on the experimental results on driven piles (e.g. Elson 1984), can also be applied to bored piles, especially at low loads. According to the numerical results of the single pile, the idealised construction sequence changes the directions of principal soil stresses at the upper part of the pile, but it does not have obvious effect on the behaviour of the laterally loaded pile. With a 30MN kentledge loading placed around the pile before the lateral load test, the pile subjected to lateral loads deflects 15% smaller than the one without kentledge loading. Furthermore, the parametric analysis of the pile group revealed that the soil axial response is found to be important in governing the lateral response of the pile group only when the ultimate shaft resistance is so low that it determines the failure mode of the group pile (i.e. tensile failure), rather than the structural capacity of the pile. Due to the reduction of pile cap effectiveness (in terms of the rotational stiffness and ultimate moment resistance), the lateral deflection increases, both group piles move upward and the location of the maximum bending moment changes from the pile head to some depth below the ground surface.