Kosslyn (1987) proposed that the left hemisphere is better than the right hemisphere at categorical visuospatial processing while the right hemisphere is better than the left hemisphere at coordinate visuospatial processing. In 134 patients, one hemisphere (and then usually the other) was temporarily deactivated by intracarotid injection of sodium amobarbital. After a hemisphere was deactivated, a cognitive test battery was conducted, which included categorical and coordinate visuospatial tasks. Using this technique, the processing capabilities of the intact hemisphere could be determined, thus directly testing Kosslyn's hypothesis regarding hemispheric specialization. Specifically, if the left hemisphere does preferentially process categorical visuospatial relationships, then its deactivation should result in more errors during categorical tasks than right hemisphere deactivation and vise versa for the right hemisphere regarding coordinate tasks. The pattern of results obtained in both categorical and coordinate tasks was consistent with Kosslyn's hypothesis when task difficulty was sufficiently high. However, when task difficulty was low, a left hemispheric processing advantage was found for both types of tasks indicating that: (1) the left hemisphere may be better at "easy" tasks regardless of the type of task and (2) the proposed hemispheric processing asymmetry may only become apparent during sufficiently demanding task conditions. These results may explain why some investigators have failed to find a significant hemispheric processing asymmetry in visuospatial categorical and coordinate tasks.