Remote operation of robot manipulators plays a crucial role in various safety-critical application areas such as forestry, mining, surveillance, etc. The capture and display of visual information of the real operation environment in an information-rich way is one of the key factors in achieving effective and robust teleoperation capability. In this paper, through subjective experiments, we comparatively investigate the roles of monocular, i.e. 2D, and stereo capture and display systems in the context of teleoperation of a robot arm. In particular, the positioning task is considered and, for both monocular and stereoscopic cases, two different modes of capture are implemented, which are static capture setup separately located from the robot arm and dynamic capture setup positioned at the tip of the robot arm. The experiments, conducted on 10 subjects (aged 24-33), show that stereo vision systems enable significant increase in accuracy of positioning compared to conventional 2D capture and display cases. In average, the tasks are completed with highest accuracy in the case of dynamic stereo capture setup.