Although previous studies indicated that the stability properties of interlimb coordination largely result from the integrated timing of efferent signals to both limbs, they also depend on afference-based interactions. In the present study, we examined contributions of afference-based error corrections to rhythmic bimanual coordination using a kinesthetic tracking task. Furthermore, since we found in previous research that subjects activated their muscles in the tracked (motor-driven) arm, we examined the functional significance of this activation to gain more insight into the processes underlying this phenomenon. To these aims, twelve subjects coordinated active movements of the right hand with motor-driven oscillatory movements of the left hand in two coordinative patterns: in-phase (relative phase 0°) and antiphase (relative phase 180°). They were either instructed to activate the muscles in the motor-driven arm as if moving along with the motor (active condition), or to keep these muscles as relaxed as possible (relaxed condition). We found that error corrections were more effective in in-phase than in antiphase coordination, resulting in more adequate adjustments of cycle durations to compensate for timing errors detected at the start of each cycle. In addition, error corrections were generally more pronounced in the active than in the relaxed condition. This activity-related difference was attributed to the associated bilateral neural control signals (as estimated using electromyography), which provided an additional reference (in terms of expected sensory consequences) for afference-based error corrections. An intimate relation was revealed between the (integrated) motor commands to both limbs and the processing of afferent feedback. © 2007 Springer-Verlag.