This paper addresses the problem of scheduling a nonpreemptive open shop with the objective of minimizing makespan. A neighborhood search algorithm based on the simulated annealing technique is proposed. The algorithm is tested on randomly generated problems, benchmark problems in the literature, and new hard problems generated in this paper. Computational results show that the algorithm performs well on all of the test problems. In many cases, an optimum solution is found, and in others the distance from the optimum or lower bound is quite small. Moreover, some of the benchmark problems in the literature are solved to optimality for the first time.