The cause of anosognosia for hemiplegia (AHP) remains unclear. Weakness is detected when there is a mismatch between the expectancy of movement and the sensory perception of movement. The feed-forward hypothesis of AHP posits that there is a failure of detection because there is a loss of motor intention and expectancy of movement. We tested motor intention by measuring the activation of proximal muscles (pectoralis majoris) while subjects squeezed a dynamometer with each hand. We tested a group of normal controls, a group of patients with hemi-paresis, a patient with neglect, a patient with resolved AHP, and a patient with persistent AHP. The patient with AHP did not contract either of his pectoralis muscles when asked to squeeze with his contralesional, paretic hand, yet he contracted both of them when squeezing the dynamometer with his ipsilesional hand. Normal controls, hemiparetic controls, and the patient with hemispatial neglect contracted both pectorales when asked to squeeze with each hand. The pattern of activation seen in the patients with persistent AHP and resolved AHP demonstrates a loss of motor intention and lends support to the feed-forward hypothesis of AHP. © 1994 American Academy of Neurology.