This study was designed to compare the relevance of self-report questionnaires and performance-based tests to assess movement imagery ability in sports. Participants included elite and novice athletes, from fencing, judo and wrestling, who completed a self-report, the Movement Imagery Questionnaire – Revised (MIQ-R; Hall & Martin, 1997), and two performance-based tests, the Movement Imagery Specific Test (MIST), and the Mental Rotation Test (MRT; Vandenberg & Kuse, 1978). There was no significant effect of the expertise variable on the MIQ-R performance, although the results yielded a positive effect of expertise on the MIST and on the MRT. Besides, results showed no correlations between the MIQ-R and the MIST, or between the MIQ-R and the MRT. However, we found a correlation between the MIST and the MRT. These findings are in line with research dissociating imagery measured by self-reports and spatial ability assessed through performance-based tests, and are discussed in terms of their implication in using self-report questionnaires in experimental psychology in general, and to assess movement imagery ability in sports in particular.