Controllability of motor imagery is one of the critical factors for the effectiveness of mental practice. It is generally defined as the ability to control or manipulate motor imagery menta1ly. In this study, a new test to measure the controllability of motor imagery (CMI test) was developed and its validity and reliability were examined. The CMI test was constructed from 15 different tasks with the following five imagery controls per one task. University students were required to image a basic movement pose and then to control or manipulate its imagery successively according to the instruction. After the five successive control, the subjects were asked to choose the last movement pose in their imagery out of five different poses (pictures) by using the recognition method. First of all, split-half and test-retest reliability estimates were obtained for the CMI test. Cron-bach'sα coefficient (an estimate of internal reliability) was 0.731 and test-retest reliability with the two months interval was 0.728. Validity of the CMI test was examined from the following three viewpoints: (1) comparison of the CMI test score between gymnastic and non-gymnastic groups; (2) relationship between the CMI test and Gordon Test of Visual Imagery Control; and (3) effect of the CMI test score on a learning sequential movement pattern. All of these results were successful in validating the CMI test. It was considered that utility of the newly developed test for controllability of motor imagery might hold sufficient expectations, though several further empirical works were needed to generalize the validity of the CMI test.