Within the limited research literature on the topic, there is considerable controversy over the usefulness of stereoscopic TV displays for performing remote manipulation tasks. Some investigators argue that a second video channel might just as well be allocated to a camera with an appropriately separated view of the worksite -- an `orthogonal'' view to that of the first camera. Other researchers argue that even though operators tend to express strong subjective preferences for stereoscopic displays, these displays often do not provide objective performance advantages. In this experiment, a group of relatively inexperienced manipulator operators performed a complex and difficult line threading task remotely and varied the visual displays available to the operators while performing this task. For each video display condition tested, the operator sat in a centered position facing two CRTs, each providing a separate view of the remote task site. Three combinations of video display types were tested: (1) monoscopic view plus orthogonal view, (2) stereoscopic view plus orthogonal view, and (3) stereoscopic view plus monoscopic view. Total task completion times, manipulative errors, and operator gaze references were measured for each combination of display types. Results show a strong and consistent operator viewing preference for stereoscopic displays as well as substantial and statistically significant performance advantages for those display combinations that provided a stereoscopic view over those that provided only monoscopic views.