Abstract Automated guidance systems for precision agriculture rely on Global Navigation Satellite Systems (GNSS) and correction services for high accuracy and precision in field operations. This study evaluates the performance of selected GNSS positioning services for precision agriculture in a field experiment. We use three correction services: SF1, SF3, and RTK, which apply varying positioning concepts, i.e., Wide Area Differential GNSS, Precise Point Positioning, and Real-Time Kinematics, respectively. The tractor is autonomously steered along multiple predefined paths located in open-sky areas as well as near the heavy tree cover. The reference route of the vehicle is determined by classical surveying. Tractor trajectories, a SF1 and SF3 corrections, are shifted from predefined straight paths, unlike in the case for RTK. Offsets of up to several decimeters are service- and area-specific, indicating an issue with the stability of the reference frame. Additionally, the varying performance of the correction services implies that environmental conditions limit the precision and accuracy of GNSS positioning in precision agriculture. The pass-to-pass analysis reveals that SF1 improves the declared accuracy, while SF3 is less reliable in obstructed areas. RTK remains a stable source for determining position. Under favorable conditions, the pass-to-pass accuracy at 95% confidence level is better than 11.5 cm, 8.5 cm, and 4.5 cm for SF1, SF3, and RTK, respectively. In the worst-case scenario, the corresponding accuracies are: 25.5 cm, 65.5 cm, and 22.5 cm.