Global Navigation Satellite System (GNSS) receivers are now widely used for navigation and speed measurements. The majority of manufacturers of chipsets and receivers claim that their products are accurate to about 0.1 km/h in measuring speed. Therefore, it is a metrological challenge to test GNSS receivers in real world environments as test vehicles are generally not capable to provide such accuracy. The datasheets for GNSS receivers, however, provide no information about specific conditions when the claimed speed accuracy might be achieved. Limited practical research activities with traceable to national standards test vehicles were conducted to reveal the practical speed accuracy parameters of GNSS receivers in different conditions. This research firstly aims to design and implement an electronic system for a test vehicle which would generate speed records with high accuracy and traceability to national standards. High accuracy of the test vehicle and synchronization of its speed records with Universal Coordinated Time allowed conducting subsequent testing of GNSS receivers of different complexities for speed. Test results demonstrate that different GNSS receivers have different speed accuracy parameters both statistically and in regards to generation of outliers. GNSS environmental factors should be taken into consideration when relying on individual speed measurements.