Accurate and reliable speed and heading angle data are essential for vehicular navigation, position reporting, cruise control, and advanced driver-assistance systems (ADAS). However, the heading angle derived based on Global Navigation Satellite System (GNSS) position information is sometimes misleading, particularly at low speed. This paper proposes the use of the sensor fusion approach to address this issue. The statistical properties of the standard speed/heading angle resolution process are analyzed, and an explicit expression of the variance is obtained. Accordingly, an adaptive two-stage filter architecture is proposed to fuse GNSS receiver and gyro measurements. This leads to a method that has the benefits of simple implementation, cost effectiveness, ease of tuning, and performance assurance.