1. The cardiorespiratory response to imagination of previously performed treadmill exercise was measured in six competitive sportsmen and six non-athletic males. This was compared with the response to a control task (imaging letters) and a task not involving imagination (‘treadmill sound only’). 2. In athletes, imagined exercise produced increases in ventilation which varied within and between subjects. The mean maximal increase (11.71 min-1) was approximately 20% of the ventilatory response to actual exercise. This was primarily due to treadmill speed-related increases in respiratory frequency (mean maximal increase, 14.8 breaths min-1) and resulted in significant reductions in end-tidal PCO2 (mean maximal fall, 7 mmHg). These effects were greater (P < 0.01) than any observed during the control tasks. 3. Changes in heart rate (mean increase, 12 beats min-1) were not significantly different from those observed during the control tasks (P > 0.2). 4. In non-athletes, imagination of exercise produced no changes in cardiorespiratory variables. No significant differences were detected in subjective assessments of movement imagery ability between athletes and non-athletes (P = 0.17). 5. This study demonstrates that ventilatory effects, when observed, are specific to imagination of exercise. The greater likelihood of generating ventilatory responses in highly trained athletes, experienced in ‘rhythmic’ sports, may be related to awareness of breathing and its role in exercise imagination strategy. A volitional component of the response cannot be discounted.