This paper presents a laser-based pedestrian tracking by multi-mobile robots in indoor environments. Each robot finds pedestrians based on an occupancy-grid-based method and tracks them by Kalman filter and Global-nearest-neighbor-based data association. It also generates a local map by EKF-SLAM. When the robots are located in vicinity, they exchange the information of tracked pedestrians and local maps through intercommunication. The tracking data are blended based on Covariance Intersection, and a global map is built by fusing the local maps so that the tracking performance can be dramatically improved. The experimental result of tracking two pedestrians by three mobile robots validates the proposed method.