One useful source of depth information available to the human nervous system is present in the horizontal disparities that exist between the two retinal images (stereoscopic depth). The relationship between horizontal disparity and depth varies with viewing distance so that an interpreting signal is required if disparities are to yield useful information. One potentially useful interpreting signal is available from ocular vergence. A number of studies have concluded, however, that a vergence signal does not provide veridical stereoscopic depth. All of these studies required observers to make a range of judgements under conditions of uncertainty (often using random dot stimuli) and we suggest that the lack of veridicality arose because of a contraction bias: a general tendency to bias judgements towards the centre of the range of possible responses. We re-examined the role of ocular vergence in the maintenance of stereoscopic depth constancy for real three-dimensional objects. Our results question the conclusions reached by previous studies and suggest that vergence can provide a veridical interpretation of stereoscopic depth. Our results indicate that horizontal retinal image disparities are not interpreted by a 'higher order' signal (i.e. the 'perceived distance' of the fixation point). The results of the experiment have significant implications for models of depth processing from disparity.