This study aims to investigate the idea that an active-resistive training with an EMG-based exoskeleton robot could be beneficial to muscle strength and antagonist muscle cocontraction control after 4-week intensive elbow flexion/extension training. Three older people over 65 years participated the training for an hour per session and completed total 20 sessions during four weeks. Outcome measures were chosen as the maximum joint torque and cocontraction ratio between the biceps/triceps brachii muscles at pre-/post-training. The Wilcoxon signed-ranks test was performed to evaluate paired difference for the outcome measures. As a result, there was no significant difference in the maximum flexion or extension torque at pre- and post-training. However, the cocontraction ratio of the triceps brachii muscle as the antagonist was significantly decreased by 9.8% after the 4-week intensive training. The active-resistive training with the exoskeleton robot in the older people yielded a promising result, showing significant changes in the antagonist muscle cocontraction.