A method for classifying an occupancy status of a vehicle seat includes associating a filtered class to a specific seat occupancy status based on a plurality of successive readings of at least one parameter value of a seat occupancy sensor, including the steps of: 1) determining an actual value of a first parameter of the seat sensor and associating an actual class to the actual value of the first parameter; 2) storing the actual class in a buffer, and deleting the oldest stored class; 3) plotting a confidence parameter indicative of the reliability of the actual class; 4) setting an updated filtered class to equal (i) an average of the classes stored in the buffer, if the confidence parameter is smaller than a first confidence threshold value, or (ii) the previously determined filter class, if the confidence parameter is greater than or equal to the first confidence threshold value.