Clinical evaluation of knee instability is often difficult to reproduce, and several different physical tests have become popular. In an attempt to elucidate reasons for variations in the degree of pivot shift phenomenon seen with the use of the various tests, we have prospectively evaluated a group of 37 patients with surgically documented ACL injuries, noting the effects of hip position and tibial rotation. The patients were examined under anesthesia, and the pivot shift was graded as 0 (absent, or negative), 0.5+ (trace), 1+, 2+, or 3+ (with locking). All knees were tested in hip abduction, neutral, and hip adduction, and with the tibia in external and internal rotation, so that six positions were evaluated. Hip position strongly correlated with the degree of pivot shift regardless of tibial rotation. Overall, abduction produced the greatest degree of pivot shift, followed by neutral and finally adduction. External tibial rotation increased the pivot shift score in abduction and neutral, but not in adduction. A grading system for the subject population showed that abduction/external rotation (ABDER) resulted in the highest pivot shift scores, and that adduction/external rotation (ADDER) and adduction/internal rotation (ADDIR) resulted in the lowest scores. Nine patients out of 20 with a 3+ pivot shift in ABDER were negative in ADDER. The pivot shift score was dampened at least one grade from ABDER to ADDER in 92% of the patients. We conclude that hip position and tibial rotation affect the degree of pivot shift phenomenon, and it is our impression that the iliotibial band plays a significant role in controlling the degree of pivot shift observed.(ABSTRACT TRUNCATED AT 250 WORDS)