The purpose of this study was to investigate the efficiency or speed of the frequently used L versus T visual search when sensory effects were controlled, ie ‘set size’ was not defined as the number of distractor patterns, but the number of distractors in the display was kept constant and the number of possible target positions varied. A search display consisted of an L-target among T-distractors, and the observer's task was to report the presence or absence of the target (experiment 1) or to identify it (whether the L-target was left-facing or right-facing; experiment 2). The observer was instructed prior to each stimulus block, about the display locations in which the target could appear. In both experiments, search time increased significantly with an increasing number of possible target locations, thus indicating that L versus T search is not ‘serial’ owing to sensory effects. Because, in the first two experiments, a search display was visible until the observer gave a response, ‘serial’ search might have resulted just from eye movements. Therefore, a control experiment was run in which display duration was limited to 150 ms. The results of this experiment showed that, even when eye movements were prevented, the search still occurred ‘serially’, ie response time increased as a function of the number of possible target positions.