In the last four to five decades piled raft foundations has been successfully applied in different parts of the world for optimization of civil engineering structure foundations. This thesis is an extension of research works being carried out to wide spread the concept of piled raft foundation further. A piled raft is a composite foundation in which the piles are used as settlement reducers and they share, with the raft, the load from the superstructure. Nevertheless, research in this area has been lagging because of the complexity of the problem and lack of field data. Numerical modeling can be used to provide valuable data with a high level of success. A three-dimensional finite-element model of a piled raft foundation was developed to simulate the case of a piled raft foundation. The model was used to examine the effect of the key parameters governing the performance of this foundation during loading and, accordingly, the load shared by the piles and the raft. After validating the numerical model with available data in the literature, the model was used to develop data for a wide range of parameters and to examine the role of the number of piles and foundation geometry, including pile spacing in the group, pile diameter, and raft thickness. From the parametric study, it is found that as pile number increases the Ultimate Load and Allowable Load capacities, Load Improvement Ratio, Settlement Ratio, Differential settlement Ratio and proportion of load carried by piles are highly improved. Diameter of pile positively affects all except differential settlement. Differentia settlement is not affected by the piled diameter. Raft thickness has significant effect on differential settlement than any other parameter. Raft thickness has no that much significant effect on other effects like Ultimate Load and Allowable Load capacities, Load Improvement Ratio, Settlement Ratio and proportion of load. Spacing considered in this thesis (3D and 4D) have no any significant effect on any of the effects mentioned. The results explained here are conclusions drawn based on the limited data found from the numerical model analysis.