In 2 experiments, observers were required to identify corresponding points on an object viewed from multiple orientations. On each trial, a surface was presented initially with a single target location marked by a small dot. Following a brief blank interval, the same surface was presented again at a different orientation. The observer was required to position an adjustable probe dot in this 2nd display to match the location of the target in the 1st view. Under optimal conditions, the variance in their settings over multiple trials was just a few minutes of arc, though these errors varied significantly with the structural complexity of the depicted surface.