Previous studies have shown that autonomy support (AS) can foster a person’s motivation and facilitate motor learning. However, the effects of AS on observational motor learning are not well understood. The present study investigated this issue by manipulating to-be-observed-model. Forty-eight male students were assigned into autonomy, yoked, and no-demonstration control groups. Three male Baseball coaches acted as models A, B, and C. Model A was instructor of students of AU group and acted as a model with high social status for AU group. Models B and C were not familiar for all participants and acted as low social status models. Participants were asked to perform a Baseball-pitch into a target during pretest (10 trials), acquisition phase (5 blocks of 10 trials), and retention test (10 trials). Prior to each acquisition block, the participants of AU and YO groups observed a model three times. Participants of AU group were free to choose model A, B, or C for any single observation. Participants in YO group were matched with those in AU group. Movement outcome, movement form, self-efficacy (SE), perceived learning effect (PLE), and perceived model attractiveness (PMA) were measured as dependent variables. Results showed that AS, relative to yoked and control conditions, led to better movement outcome during acquisition and retention. Action observation enhanced movement form during acquisition and retention. AS increased SE, PLE, and PMA compared with yoked and control conditions. Results provide support for the OPTIMAL theory and indicate that AS facilitates observational motor learning.