The purposes of the present study were to compare knee flexion torques determined in three test positions: (1) sitting; (2) prone; and (3) standing. Twenty-one healthy males (mean age 23 +/- 2 years) completed tests in all three positions on each of two occasions, using isolated concentric and eccentric muscle actions. All tests were completed at 60 o /sec angular velocity, using an isokinetic dynamometer, with the sitting and prone tests completed using the dynamometer chair/bed. A specially built standing/stabilization frame was constructed to test knee flexion in standing. Averaging scores over two occasions was required to achieve reliability coefficients (ICCs) greater than 0.75 for all tests. Standing tests were characterized by the lowest ICCs and the largest measurement error. Repeated measures analysis of variance (three positions by two muscle actions) indicated that the sitting position produced higher torques than did the prone (P < 0.01) and standing positions (P < 0.01), and the standing position produced higher torques than did the prone position (P < 0.01). All 21 subjects identified the standing position as permitting the fewest accessory movements, and providing the greatest isolation of the knee flexion movement and the hamstring muscles. Standing scores should not be compared with those determined in sitting or prone positions. In situations where the clinician wishes to isolate the knee flexors, the standing position is practical and may be advantageous. Future research needs to determine if reliability for standing tests can be improved and to examine the strength-function relationship for all three positions for testing knee flexion strength.