According to Kohl and Roenker (1983, JMB, 15, 179-190), inflow mechanisms presume that neuromuscular activity during imagery gives peripheral feedback that facilitates appropriate conceptualization for future imagery trials.  By appropriate, it is meant that the pattern of neuromuscular activation during imagery is the same, but reduced, relative to the pattern observed when performing the actual task.  This would thus constitute a basis for correct feedback from peripheral sources and an explanation for how imagery improves skill learning and performance.  As a test of the inflow mechanism, the purpose of this study was to establish: (a) whether there are localized increases in muscular activation during imagery; and (b) whether the pattern of increased activation during imagery mirrors the muscular activity observed during real movements of the task being imagined.  Undergraduate students (N = 60) completed a series of real and imagined sagittal plane dumbbell and tansverse plane manipulandum curls.  For both movements, biceps and triceps EMG activity was measured during baseline, imagery, and actual movement conditions.  Average EMG biceps and triceps activity was significantly greater in the active arm during imagery than during baseline (ps < 0.05).  However, pattern analysis showed that the activation did not mirror the pattern observed during actual movement.  Results do not support theories that depend on inflow mechanisms (e.g., psychoneuro- muscular theory) since the pattern of increased activation during imagery did not reflect that observed during actual movement.  These results suggest that more attention should be given to theories that rely on outflow mechanisms.