This paper extends previous work (Gandevia & McCloskey, 1976) on proprioception in the terminal joint of the middle finger. By positioning the finger in appropriate ways proprioceptive acuity at the joint can be assessed when no muscular afferents could contribute, or when afferents in the flexor but not the extensor could contribute, or when afferents from both muscles could contribute. Digital nerve block anaesthetizes joint and cutaneous receptors and so was used to study the contributions from muscle afferents in isolation. Displacements (10 degrees) at various angular velocities were better detected when muscle afferents from both flexor and extensor muscles could contribute. This was so whether joint and cutaneous receptors were also available, or after digital anaesthesia. Performance when only muscle afferents are available is, however, inferior to that when all sensory mechanisms are intact. It is concluded that muscle afferents contribute to kinaesthesia, and that a full complement of such receptors from agonist and antagonist muscles gives superior acuity to that achieved when only the receptors of one of the muscle groups is available. The angular displacements necessary for 70% correct detection were determined at angular velocities between 0.25 degrees and 160 degrees/s. Proprioceptive performance was optimal with all proprioceptive mechanisms intact over the range of angular velocities 10 degrees -80 degrees/s: 70% correct detection of displacements of 0.8 degrees-1.2 degrees occurred in this range. Performance deteriorated slightly at higher velocities of displacement. Performance was significantly poorer when only joint and cutaneous receptors could contribute (in the absence of intramuscular receptors), and when only intramuscular receptors could contribute (in the absence of joint and cutaneous receptors). Full proprioceptive acuity depends upon the availability of receptors in muscles and in skin and/or joints.