A structure optimization for the initial design stage of ship using FEM is considered in this paper. A general bulk carrier is selected as the target ship. Both the shape of the bottom structure and the size of plate thickness of the bulk carrier are taken as design variable simultaneously for a weight minimal optimization problem. Five geometric dimensions and thirty-one plate thicknesses are taken as the design variables for the shape and the size of the bottom structure, respectively. The yield stress for all elements of the three hold FEM model and the buckling stress for the selected stiffened panels are evaluated as the constraint conditions. The selected stiffened panels are on the bottom plate, the inner bottom plate, the girder plate, the floor plate, and the bilge hopper plate. The optimal calculations are performed with five combinations of the design variables and the constraint conditions, and the designs obtained by the optimal analyses are compared. It is shown that the structural optimization with the design variables for the shape and size of the bottom structure can reduce the weight of ship structure and satisfy the constraint conditions effectively. Since the design variables take discrete values, this problem becomes combinatorial optimization. So, the genetic algorithm is used for the optimal analyses.