When attempting to determine the middle of a line, patients with neglect deviate from true center. Deviation may be induced by perceptual-attentional bias, premotor-intentional bias, or both. Using a video-based apparatus, we decoupled perceptual from premotor influences on line bisection performance in patients with hemispatial neglect to examine (a) the relationship between primary and secondary bias and (b) the relationship of bias type to lesion location. The same video-based procedure was applied to target cancellation to determine if neglect type varied as a function of task. Primary attentional-perceptual bias was found using line bisection in 14/26 subjects, most of whom had lesions involving the posterior hemisphere. Primary premotor-intentional bias on line bisection was more often associated with lesions of frontal-subcortical structures. The neglect type determined by the bisection task agreed with the results of target cancellation in most cases. Secondary bias was determined based upon whether decoupling decreased the magnitude of bisection error (concordant), increased error (discordant), or produced no significant change. Most patients showed a secondary bias, with 12/26 in the discordant group and 11/26 in the concordant group. Discordant secondary bias was more common in premotor-intentional neglect (10/12) than in perceptual-attentional neglect (2/14), whereas concordant bias was more common in the latter group (10/14) compared to the former (1/12). The nonrandom relationship between primary and secondary bias may provide a more detailed description of ways in which anatomically separate components of a cortical network contribute to spatial processing under conditions of perceptuomotor incongruity.