The use of navigation systems, such as those based today on the Global Positioning System (GPS), in standard vehicles has become affordable and has been generalized, in the recent years. Combining the knowledge of our position with our heading can bring an added value to many applications, as has been seen in many smartphones that nowadays come equipped with a GPS and a magnetic compass. This paper presents an algorithm for a digital compass, with heading estimation based on positioning data retrieved from two low-cost single frequency GPS receivers. The algorithm is based on an Extended Kalman Filter (EKF) that was developed for heading estimation.The performance of the proposed algorithm was evaluated using real data from two low-cost receivers installed in a standard car. Several field trials were conducted and three motion scenarios were analyzed: straight path, curved path and immobilization.