A conventional picture viewed from its center of projection (CoP) produces the same retinal image as the original depicted scene. When viewed from another position, the retinal image specifies a different scene, but people typically perceive the original depicted scene, not the one specified by the retinal image. Thus, they compensate for incorrect viewing position. Compensation is based on a measurement of the slant of the picture surface, and the primary cue for the measurement is binocular disparity. Using disparity works because disparity signals the slant of the picture surface and is unaffected by the depicted 3d contents of the picture. In stereograms, the disparity signals both the 3d contents of the depicted scene and the slant of the picture surface. Here we ask if observers compensate for incorrect viewing position with stereograms in a fashion similar to the compensation with conventional pictures. We conducted a series of experiments in which observers varied the dihedral angle of a stereoscopic hinge stimulus. Observers indicated whether the dihedral angle was greater or less than 90 deg. The stimulus was presented on a computer display, which could be rotated about the vertical axis to vary obliqueness of viewing position. We compared the perceived right angle to that predicted if observers did or did not compensate for oblique viewing angle. We also manipulated the information the observer had about viewing position from minimal (isolated hinge stimulus, display frame invisible) to maximal (hinge stimulus embedded in conventional picture, display frame visible). We observed little evidence for compensation in all conditions. This suggests that the compensation mechanism for incorrect viewing position that is used with conventional pictures is not used with a disparity-specified scene in stereograms.