Abstract
A drop algorithm is proposed for solving the p median problem. In solving for a fixed value of p, tight bounds on all other median solutions in the range m − 1 to p + 1 are generated where m is the number of possible location sites and p is the number of medians. A step by step numerical example is described, and extensive computational experience is reported for some standard sets of tests problems in the literature. Comparisons with the well-known greedy interchange heuristic confirm the effectiveness of this drop approach in solving a significant number of difficult median problems.