Automatic Identification System (AIS) data are prone to alterations that impede signal storage, producing incomplete trajectories. Since trajectory analysis assumes complete AIS data sets, incomplete trajectories are discarded. Yet, those data sets contain proper values, which could be exploited. This paper describes an approach to estimate the degree of reliability assigned to missing segments of a vessel trajectory, depending on meaningful statistical variations of recorded and predicted vessel positions. The approach was tested on real AIS data of three vessels. Results suggest that reliability can be determined from vessel speed and heading variations.