An approach to solving general linear 0-1 restricted optimization problems on a digital computer is presented.The approach used employs a combination of partial and random search conducted over the space of 2 ' ~ n-dimensional binary vectors.The partial search moves from randomly selected starting points, through the space, to precisely characterized local optima.Random search is then reapplied in an attempt to overcome the problem of generating a locally optimum solution which is not globally optimum.The resulting algorithm has been implemented on the GE 635 computer.Computational experience obtained thus far indicates that the approach taken is feasible for solving such problems and offers the possibilities of obtaining good solutions to large problems at modest computation costs.