Summary: A spatial co-location is a set of spatial events being frequently observed together in nearby geographic space. A common framework for mining spatial association patterns employs a level-wised search method (like Apriori). However, the Apriori-based algorithms do not scale well for discovering long co-location patterns in large or dense spatial neighborhoods and can be restricted for only short pattern discovery. To address this problem, we propose an algorithm for finding maximal co-located event sets which concisely represent all co-location patterns. The proposed algorithm generates only most promising candidates, traverses the pattern search space in depth-first manner with an effective pruning scheme, and reduces expensive co-location instance search operations. Our experiment result shows that the proposed algorithm is computationally effective when mining maximal co-locations.