Proprioception was quantified in a group of patients who had documented complete ACL tears. Threshold to detection of passive change in position of the knee was measured using a well-described test. Eleven pa tients with arthroscopically proven complete ACL tears and findings consistent with moderate to severe anter olateral rotatory instability were tested. Testing was done within the 30° to 40° range of knee flexion. Patients were blindfolded and the injured and uninjured knees were tested in random sequence so that the normal knee could serve as an internal control. Testing was also done in a blind manner, i.e., the examiner did not know which knee had been injured. An age- matched control group underwent identical testing. Po tentially significant variables such as age, time from injury, and degree of rehabilitation as measured by thigh circumference and isokinetic testing of the knee were included in a multivariate analysis. Control subjects demonstrated virtually identical threshold values between their two knees, the mean variation being less than 2%. The test group, however, showed a significantly higher mean threshold value for the injured versus the noninjured knee (P < 0.01), the mean variation being over 25%. Multivariate analysis demonstrated that changes recorded in the propriocep tion of the injured knee were attributable to the loss of the ACL rather than to other variables. Patients who have complete ACL tears and moderate to severe rotatory instability may also experience a decline in proprioceptive function of their knee.