Either for recreational or professional reasons, ships travel across the globe generating a network of maritime traffic with routes connecting key areas such as ports, off-shore facilities or fishing areas. Monitoring vessels’ position relatively to maritime routes provides crucial information about their destination, and can help reducing the risk of collision. In this paper, we implement a fuzzy logic approach for associating vessels to maritime routes, suitable to variable surveillance contexts and very sparse data. Moreover, the framework is agnostic to the way maritime routes are provided, either reflecting patterns-of-life from statistical models extracted from real data or being hand-crafted by a user. Fuzzy membership functions enable expressing that vessels can belong more or less to route corridors, while they follow only one of the possible routes. The computation of membership scores relies on a precise distance computation involving geometrical properties of Earth, valid for very large route segments. The defuzzification step allows non-specific outputs. Several instantiations with aggregation operators of different semantics are compared on a real dataset of tracklets from the Automatic Identification System, with ground truth labels of routes. The performance is assessed in a quality space along with the three dimensions of correctness, specificity and confidence.