Abstract: An improved bat algorithm with Gauss mutation and Lévy Flights (GMBA) is proposed in this paper, in which each bat chooses different flight strategy according to its state at present. Bat in worse position choose to Lévy flight behaviors.Bat in better solution move to best solution. Moreover, when meeting the condition of the variation, The GMBA performed the Gauss mutation operation to improve the ability of the bat trapping out of the local optima. Fourteen typical experiments show that the new algorithm (GMBA) is superior than BA and LBA.