Automatic Vehicle Location (AVL) systems are becoming increasingly common, especially for fleet monitoring/management applications such as probe vehicle operation. A large percentage of AVL systems report position and sometimes velocity on a periodic basis to a base station. However, sending data on a periodic basis can be costly and inefficient. It is more efficient to send data on an as-needed, aperiodic basis. It is possible to use real-time filtering algorithms to send trajectory data on an aperiodic basis using information on the spatial aspects of the roads and velocity of the vehicle. In many cases, it is only necessary to send new information when there is significant deviation both spatially and temporally. Further, velocity changes can trigger data transmissions. As part of the research described in this paper, aperiodic filtering techniques have been developed and applied to several AVL applications. Using these filtering algorithms, it is possible to decrease communications cost while improving route representation and predicted time of arrival. To test the effectiveness of the algorithms, an experimental AVL system was designed and developed. Further, the filtering techniques have been applied to large representative vehicle activity data sets, successfully reducing the communications and storage costs by 80 to 90%.