We investigated the precision of binocular gaze control while observers performed a high-precision man- 24ual movement, which involved hitting a target hole in a plate with a hand-held needle. Binocular eye 25movements and the 3D-position of the needle tip were tracked. In general the observers oriented their 26gaze to the target before they reached it with the needle. The amplitude of microsaccades scaled with 27the distance of the needle tip. We did not find evidence for the coordination of version and vergence dur- 28ing microsaccades which could be expected if those movements displaced gaze between the needle and 29the target hole. In a control experiment observers executed small saccades between marks on a slanted 30plane. Even when the observers executed saccades as small as the microsaccades in the needle experi- 31ment, we observed a coordinated displacement of the point of gaze on the horizontal and depth axis. 32Our results show that the characteristics of eye movements such as the frequency and amplitude of 33microsaccades are adapted online to the task demands. However, a coordinated control of version and 34vergence in small saccades is only observed if a movement of gaze on a slanted trajectory is explicitly 35instructed.