We investigated the effects of intermittent long-term stretch training (5 weeks) on the architectural and mechanical properties of the muscle–tendon unit (MTU) in healthy humans. MTU’s viscoelastic and architectural properties in the human medial gastrocnemius (MG) muscle and the contribution of muscle and tendon structures to the MTU lengthening were analyzed. Ten healthy volunteers participated in the study (four females and six males). The passive stretch of the plantar flexor muscles was achieved from 0° (neutral ankle position) to 25° of dorsiflexion. Measurements were obtained during a single passive stretch before and after the completion of the stretching protocol. During the stretch, the architectural parameters of the MG muscle were measured via ultrasonography, and the passive torque was recorded by means of a strain-gauge transducer. Repeated-measure ANOVA was applied for all parameters. When expressed as a percentage for all dorsiflexion angles, the relative torque values decreased (p < 0.001). In the same way, architectural parameters (pennation angle and fascicle length) were compared for covariance and showed a significant difference between the slopes (ANCOVA p < 0.0001 and p < 0.001, respectively) suggesting a modification in the mechanical behavior after stretch training. Furthermore, the values for passive stiffness decreased (p < 0.05). The maximum ankle range of motion (ROM) (p < 0.01) and the maximum passive torque (p < 0.05) increased. Lastly, the contribution of the free tendon increased more than fascicle elongation to the total lengthening of the MTU (ANCOVA p < 0.001). Our results suggest that five weeks of intermittent static stretch training significantly change the behavior of the MTU. Specifically, it can increase flexibility and increase tendon contribution during MTU lengthening.