The gate assignment problem (GAP) is an important task in airport management. This study investigates an original probability learning based heuristic algorithm for solving the problem. The proposed algorithm relies on a mixed search strategy exploring both feasible and infeasible solutions with the tabu search method and employs a reinforcement learning mechanism to guide the search toward new promising regions. The algorithm is compared with several reference algorithms on three sets of real-world benchmark instances in the literature. Computational results show the high competitiveness of the algorithm in terms of solution quality and computation time. Especially, it reports improved best solutions (new upper bounds) for all the 180 tested real-world benchmark instances in the literature. The key components of the algorithm are analyzed. The code of the algorithm will be publicly available.