Previous studies have evaluated reliability and validity of clinical measurements of femoral and tibiofibular torsion, however, the results are conflicting, warranting an alternative method. The authors developed modified methods of measuring femoral and tibiofibular torsion using different reference axes, and validity and reliability of the modified and existing methods were examined in healthy subjects. Reference values were established using magnetic resonance imaging (MRI) for the validity study. There were significant differences between all clinical methods and corresponding MRI outcomes (P < 0.05), whereas Pearson's correlation coefficients between the clinical methods and MRI ranged from 0.67-0.92. For modified methods, intraclass correlation coefficients were 0.89-0.97, whereas those of existing methods were 0.75-0.81. Modified methods were more precise, with standard error of measurements (SEM) of 1.5-2.9°, compared to existing methods (SEM = 4.0-5.1°). These results indicate that clinical methods available currently do not accurately measure true torsion of the femur and tibia. Nevertheless, the clinical methods may be useful for screening and descriptive purposes as indices of true torsion, and use of different reference axes could improve their reliability.