This paper investigates a two-stage model of infants' imitative learning from observed actions and their effects. According to this model, the observation of another person's action activates the corresponding motor code in the infants' motor repertoire (i.e. leads to motor resonance). The second process guiding imitative behavior results from the observed action effects. If the modeled action is followed by a salient action effect, the representation of this effect (i.e. perceptual code) will be associated with the activated motor code. If the infant later aims to obtain the same effect, the corresponding motor program will be activated and the model's action will therefore be imitated. Accordingly, the model assumes that for the imitation of novel actions the modeled action needs to elicit sufficient motor resonance and must be followed by a salient action effect. Using the head touch imitation paradigm, we tested these two assumptions derived from the model. To this end, we manipulated whether the actions demonstrated to the infants were or were not in the motor repertoire, i.e. elicited stronger or less strong motor resonance, and whether they were followed by salient action effects or not. The results were in line with the proposed two-stage model of infants' imitative learning and suggest that motor resonance is necessary, but not sufficient for infants' imitative learning from others' actions and their effects.