Background: Many variables influence the tactile bisection performance of normal subjects. Of these, studies that investigated the starting point effect have reported inconsistent results. Objective: To determine if the final search direction rather than the initial search direction (starting point) may have an effect on tactile bisection performance, especially when subjects are making multiple searches. Methods: The authors divided the experiment into single- and multiple-search tasks. Thirty-two blindfolded normal subjects were asked to indicate the midpoint of a rod after a one-way search (single-search task) or after two or more one-way searches (multiple-search task). In each task, the subject started on either end of the horizontally placed rods. The same procedure was also conducted with rods oriented vertically or radially. Therefore, there were six conditions (two starting points x three orientations). Results: In the single-search task, there was a significant effect of starting point or movement direction in all six conditions, with biases occurring toward the starting point from the true midpoint (this is termed the "overshoot phenomenon"). In the multiple-search task, however, there was no significant effect of the starting point in all six conditions. Rather, biases depended on the final movement direction in five of the six conditions, occurring toward the starting point of the final search. Conclusion: Future research is needed to understand the mechanism of the overshoot phenomenon in tactile bisection by normal subjects and how the overshoot phenomenon influences patients with neglect.