Observational learning (OL) allows an individual to acquire novel responses by observing others' behavior and the corresponding consequences. The complexity of skills involved with OL may vary with the learning context. A learner may observe modeled responses to both trained (known) and untrained (unknown) stimuli or they may observe both reinforced (correct) and nonreinforced (incorrect) responses. The purpose of this study was to develop assessment and training procedures for OL component skills when the learner observes a combination of learning contexts: reinforced and nonreinforced responses to both trained and untrained stimuli. Two children with autism, Tom and David, participated. We assessed the following component skills in the context of tact trials: (1) Discriminating trained and untrained stimuli, (2) attending to the modeled performance, (3) discriminating consequences, and (4) conditionally responding based upon a name call. Next, we trained the component skill(s) for which the learner's performance did not meet criterion and then reassessed for OL. For both participants, immediate increases in OL were observed; however, modifications to the post-assessment (differential observing response for consequences and/or differential reinforcement) were required to produce (Tom) or maintain (David) criterion levels of responding. Interpretations of these outcomes, as well as limitations and directions for …