Two separate experiments examined user performance and viewer discomfort during virtual precision alignment tasks while viewing a stereoscopic 3D (S3D) display. In both experiments, virtual camera separation was manipulated to correspond to no stereopsis cues (zero separation), several levels of microstereopsis (20, 40, 60, and 80%), and orthostereopsis (100% of interpupillary distance). Viewer discomfort was assessed before and after each experimental session, measured subjectively via self-report on the Simulator Sickness Questionnaire (SSQ). Objective measures of binocular status (phoria and fusion ranges) and standing postural stability were additionally evaluated pre and postsessions. Overall, the results suggest binocular fusion ranges may serve as useful objective indicators of discomfort from S3D viewing, perhaps as supplemental measures to standard subjective reports. For the group as a whole, the S3D system was fairly comfortable to view, although roughly half of the participants reported some discomfort, ranging from mild to severe, and typically with the larger camera separations. Microstereopsis conferred significant performance benefits over the no-stereopsis conditions, so microstereoscopic camera separations might be of great utility for non-critical viewing applications. However, performance was best with near-orthostereoscopic or orthostereoscopic camera separations. Our results support the use of orthostereopsis for critical, high-precision manual spatial tasks performed via stereoscopic 3D display systems, including remote surgery, robotic interaction with dangerous or hazardous materials, and related teleoperative spatial tasks.