This study investigated the relationship of controllability of mental imagery with canoe-slalom performance. Controllability of mental imagery was assessed by an objective test of mental rotation, the Mental Rotations Test. This test was administered to both elite (n = 19) and intermediate (n = 12) athletes. Predictive validity of the controllability test was supported by a significant correlation between test scores and race rank-order for the elite canoeing group (rs = 0.42, p<.05); however, it did not distinguish elite from intermediate groups (t29 = 0.98, p>.05). Researchers should attempt to evaluate vividness of imagery, controllability of imagery, and accuracy of reference to understand more fully the nature of athletes' imagery.