Numerical Algorithms 35: 155–173, 2004.  2004 Kluwer Academic Publishers. Printed in the Netherlands. A computational study of global algorithms for linear bilevel programming Carlos Henrique Medeiros de Sabóia a,∗ , Manoel Campêlo b,∗∗ and Susana Scheimberg c,∗∗∗ a COPPE/Sistemas, Universidade Federal do Rio de Janeiro – UFRJ, Brazil b Dep. Estatística e Mat. Aplicada, Universidade Federal do Ceará – UFC, CEP 60455-760, Fortaleza-CE, Brazil E-mail: mcampelo@ufc.br. c DCC/IM, COPPE/Sistemas, Universidade Federal do Rio de Janeiro – UFRJ, C.P. 68511, CEP 21945-970, Rio de Janeiro-RJ, Brazil E-mail: susana@cos.ufrj.br. Received 12 December 2001; accepted 6 August 2002 We analyze two global algorithms for solving the linear bilevel program (LBP) problem. The first one is a recent algorithm built on a new concept of equilibrium point and a modified version of the outer approximation method. The second one is an efficient branch-and-bound algorithm known in the literature. Based on computational results we propose some modifi- cations in both algorithms to improve their computational performance. A significant number of experiments is carried out and a comparative study with the algorithms is presented. The modified procedures has better performance than the original versions. Keywords: bilevel linear programming, global optimization, branch-and-bound, exact penalty methods AMS subject classification: 65K05, 90C26 1. Introduction Bilevel programming problems arise when a two-level hierarchical system is mod- elled. In this system there is a leader that makes the decision first, and a follower that makes its decision based on the leader’s. This kind of problem has a wide field of ap- plications including structural design [16], urban transportation [11], traffic assignment [22], pollution control [2], economics [14], etc. In this article the linear bilevel program is studied. It can be formulated as follows: ∗ Partially supported by CAPES at COPPE/UFRJ. ∗∗ Partially supported by CNPq, grant 300251/00-9(RE). ∗∗∗ Partially supported by CNPq, grant 302393/85-4(RN).