The purpose of this study is to clarify the best predictor for estimating the joint torque capacity of dorsiflexors muscles. Nineteen male college students (19.9±1.2 yrs) participated in this study. The anatomical cross-sectional area (ACSA) of the tibialis anterior (TA) muscle was estimated by using magnetic resonance imaging (MRI) every 0.5cm from the proximal end of the tibia to the lateral malleolus. B-mode ultrasonic images of the TA were obtained by using a 7.5MHz probe, and the fascicle angle between superficial aponeurosis and central aponeurosis of the TA was measured as the pennation angle at proximal 30%, 40% and 50% positions of the leg as well as at the maximal ACSA (ACSAmax) position. The physiological cross-sectional area (PCSA) of the TA was determined by dividing muscle volume (MV) by fascicle length, multiplied by the cosine of the angle of fiber pennation. Maximal isometric dorsiflexion torque (TQ) was measured by using an isokinetic dynamometer. ACSAmax was found at the 33.8% position on average, and ACSAmax was significantly correlated with TQ (P<0.05). At the 30% and ACSAmax positions, there were significant correlations between pennation angle and muscle thickness (P<0.01, respectively), and between pennation angle and muscle strength of ankle dorsiflexion (P<0.05, respectively). The correlation with TQ was slightly higher for PCSA (r=0.743) and MV (r=0.625) than for ACSAmax (r=0.542), but there were no significant differences. The results suggested that ACSAmax has adequate precision for estimating torque-producing capacity of dorsiflexors muscles.