Numerous studies have investigated torque-angular velocity relation of human muscles in vivo, by using isokinetic strength measurements. However, the experimental data in the low-torque, high-velocity region are typically missing, primarily because of large inertial mass and limited range of motion in the human limbs. PURPOSE: To describe the whole picture of torque-angular velocity relation of human dorsiflexors in vivo, by combining isokinetic strength measurement with the slack test, a method for determining unloaded shortening velocity of muscle (Edman, 1979; Sasaki & Ishii, 2005). METHODS: Fifteen healthy students (age: 19.5 ± 0.8 years, height: 164.4 ± 9.1 cm, body mass: 53.9 ± 8.4 kg) participated in this study. In the slack-test measurement, the quick release of ankle joint was applied while the subjects exerted an isometric torque at 15, 50, and 85% of maximal voluntary contraction (MVC). Since the relation between the angular displacement of quick release (ÄL) and the time required to redevelop the torque (Ät) was approximately linear, the unloaded shortening velocity of dorsiflexors was determined as the linear regression slope of ÄL-Ät relation. In the isokinetic measurement, peak torque was determined at angular velocities of 0, 0.26, 0.52, 1.05, 2.09, and 3.14 rad s−1. Surface electromyographic activities (EMGs) of tibialis anterior (TA) and soleus (SOL) muscles during contraction were also examined. RESULTS: The results from the slack test showed that the unloaded shortening velocity of dorsiflexors increased with the contraction intensity (R2 = 0.37, P < 0.0001), suggesting an additional recruitment of fast motor units at higher contraction intensities (i.e. 50 and 85% of MVC). When combined with the unloaded shortening velocity determined at 85% of MVC (13.6 ± 9.1 rad s−1), the isokinetic torque-angular velocity relation had a hyperbolic shape, in which the shape parameter ranged between 0.09 and 1.38. In addition, EMGs of TA and SOL during contraction were consistent across different contraction velocity (Friedman test: P < 0.24, TA; P < 0.26, SOL). CONCLUSION: Torque-angular velocity relation of human dorsiflexors in vivo had hyperbolic shape, in which the shape parameter was consistent with the other muscles of humans and animals.