Breakage or loosening of locking screws may impair fracture fixation or bone healing in locked nailing of tibial fractures. Bending strength and bone holding power, two important design objectives of locking screws, may conflict with each other. The present study used multiobjective optimization with a genetic algorithm to investigate the optimal designs with respect to these two objectives. Three-dimensional finite element models for analyzing bending strength and bone holding power of locking screws were created first. Through use of a Taguchi L-25 orthogonal array, two objective functions were developed by least-squares regression analyses. Then, the trade-off solutions between the two objectives known as Pareto optima were explored by a weighted-sum aggregating approach under geometric constraints. The objective functions, reliably reflecting the finite element results, were valid for multiobjective studies. The Pareto fronts of the screws with 4.5-mm and 5.0-mm outer diameters were similar. The "knee" region of the Pareto front, characterized by the fact that a small improvement in either objective will cause a large deterioration in the other objective, might be the favored choice of optimal designs. The commercially available locking screws compared with the Pareto optima were found to be dominated designs and could be improved. In conclusion, the multiobjective optimization with a genetic algorithm was useful for optimization of locking screw design with many variables and conflicting objectives. Choosing an optimal design requires a thorough knowledge of the inherent problems. This method could reduce the time, cost, and labor associated with the screw development process.