Across five experiments, we investigated the parameters involved in the observation and in the execution of the action of lifting an object. The observers were shown minimal information on movements, consisting of either the working-point displacement only (ie two points representing the hand and object) or additional configural information on the kinematics of the trunk, shoulder, arm, forearm, and hand, joined by a stick diagram. Furthermore, displays showed either a participant's own movements or those of another person, when different weights were lifted. The participants' task was to estimate the weight of the lifted objects. The results revealed that, although overall performance was not dependent on the visual conditions (working point versus stick diagram) or ownership conditions (self versus other), the kinematic cues used to perform the task differed as a function of these conditions. In addition, the kinematic parameters relevant for action observation did not match those relevant for action execution. This was confirmed in experiments by using artificially altered movement samples, where the variations in critical kinematic variables were manipulated separately or in combination. We discuss the implications of these results for the roles of motor simulation and visual analysis in action observation.