In this paper we have presented the basic wavelength division multiplexing network design problem with fuzzy demands. In order to solve the problem, two operative integer linear models and three metaheuristic algorithms have been described. The computational experiments carried out on both real and artificial instances corroborate that the metaheuristic algorithms perform correctly on the Fuzzy Integer Programming models and provide the managers with a fuzzy solution that makes them able to foreseen the consequences of the sudden increment of the demands.