Abstract: This paper presents a cooperative search strategy based on multiple mobile robots to realize the active olfaction (also called odour/gas source localization or chemical plume tracing). Ant colony algorithm (ACA) combined with upwind search is used to coordinate swarm robots' motion. The ACA can effectively make most robots move toward the high-pheromone area and guarantee the distances among robots not too far. The upwind search can decrease the probability of robots trapping in local optimization too early. To judge the correct motion direction, the history pheromone is also considered in the ACA. The method of combining high-concentration maintenance criterion and fluid mass flux divergence calculation is put forward to recognize the odour/gas source. Computer simulations show that the proposed search strategy can work in turbulent environments. Using this strategy, the robots can escape from both local high concentration and eddy areas. In addition, the odour/gas source can be correctly declared finally.