Previous research has shown that speakers gesture more when describing imagistic information than when describing non-imagistic information. One explanation for this finding is that spatial information is more difficult to describe verbally than non-spatial information. In the present study, we designed two novel tasks, one verbal and one spatial, in which difficulty and spatiality were not confounded. In Experiment 1, we demonstrated that the spatial task is actually less difficult and leads to fewer errors than the verbal task. In Experiment 2, we found that speakers produced more representational gestures on the spatial task than the verbal task, even though it was not more difficult. Results suggest that speakers do not gesture on spatial tasks simply because they are more difficult.