In this paper, we discuss computational aspects of an interior-point algorithm [1] for indefinite quadratic programming problems with box constraints. The algorithm finds a local minimizer by successively solving indefinite quadratic problems with an ellipsoid constraint. In addition, we present a sufficient condition for a local minimizer to be global, and we use this result to generate test problems with a known global solution. The proposed algorithm has been implemented on an IBM 3090 computer and tested on a variety of dense test problems, including problems with a known global optimizer.