These experiments were designed to test the idea that, in a forearm position-matching task, it is the difference in afferent signals coming from the antagonist muscles of the forearm that determines the perceived position of the arm. In one experiment, flexor and then extensor muscles of the reference arm were conditioned by isometric voluntary contractions while the arm was held at the test angle, approximately 45° from the horizontal. At the same time, indicator arm flexor muscles were contracted while the arm was flexed, or extensors were contracted while it was extended. After an indicator flexor contraction, during matching, subjects made large errors in the direction of flexion, by 9.3° relative to the reference arm and after an indicator extensor contraction by 7.4° in the direction of extension. In the second experiment, with reference muscles conditioned as before, slack was introduced in indicator muscles by a combination of muscle contraction and stretch. This was expected to lower levels of afferent activity in indicator muscles. The subsequent matching experiment yielded much smaller errors than before, 1.4° in the direction of flexion. In both experiments, signal levels coming from the reference arm remained the same and what changed was the level of indicator signal. The fact that matching errors were small when slack was introduced in indicator muscles supported the view that the signal coming from reference muscles was also small. It was concluded that the brain is concerned with the signal difference from the antagonist pair of each arm and with the total signal difference between the two arms.