Thresholding is an important approach for image segmentation and it is the first step in the image processing for many applications. Segmentation is a low level operation that can segment an image in nonoverlapping regions. The optimal thresholds are found by maximizing Kapur's entropy-based thresholding function in a grey level image. However, the required CPU time increases exponentially with the number of desired optimal thresholds. In this paper a global multilevel thresholding algorithm for image segmentation is proposed based on the Bat inspired algorithm (BA). Cuckoo search (CS) algorithm was also implemented and compared with Kapur’s and BA’s algorithms. All algorithms have been tested on four sample images and experimental results show that both metaheuristics find excellent solutions, while computational time is negligible compared to exhaustive search. Key-Words: Bat algorithm, Maximum entropy thresholding, Image thresholding, Optimization metaheuristics, Nature inspired metaheuristics, Swarm intelligence