AbstractA heuristic algorithm for solving capital budgeting problems is described. The algorithm proceeds from an initial solution, obtained with the use of ranking procedures, and progressively improves the solution through exchange operations. Computational experience in solving a set of 30-variable, 5-constraint problems is described and the approach to optimal values with increasing computation time is shown. The results show considerable time advantage over 0–1 implicit enumeration techniques. In only two of twenty-eight problems were the solutions nonoptimal, and in these two cases the objective function values for the heuristic solutions were within 0.5 percent of those for the optimal solutions. A brief investigation was conducted on larger problems (60 variables, 5 constraints); the time advantage is even greater on problems of that size. Details of the problems and their optimal solutions are available to interested readers upon request to the author.