The impact of goal salience and verbal cues given by the model on 3- to 5-year-olds' reproduction of action components (movement or goal) was investigated in an imitation choice task. Preschoolers watched an experimenter moving a puppet up or down a ramp, terminating at one of two target objects. The target objects were either differently colored plastic bowls (low-salient group) or sociofunctionally relevant objects (high-salient group). While demonstrating the action in several trials, the model verbalized either the movement, the goal, both the movement and the goal, or none of the components. Children imitated the action on a second ramp with reversed positions of the target objects, so they needed to decide whether to reproduce the observed movement or the observed end state. In the high-salient group, preschoolers preferentially imitated the goal component, whereas in the low-salient group, they did not show a preference for one of the components. Across trials, preschoolers preferentially imitated the goal when this component or both components were verbalized, whereas they showed no preference when the movement or none of the components was emphasized. Yet, verbal cues seemed to have stronger effects in the low-salient condition. We conclude that sociofunctional salience of action goals and communicative cues of the model affect young children's representation of observed actions, leading to a selective reproduction of those action components that are relevant to the child. (Contains 1 table and 3 figures.)