Diminished social motivation constitutes one of the core impairments of Autism Spectrum Disorder (ASD) and is thought to have a strong impact on the way individuals with autism respond to the presence of others. In this study, we hypothesized that experimental contexts involving direct interaction with an experimenter might elicit different reactions in ASD and thus act as a potential confound in the interpretation of group differences during social cognitive tests. Following classic work in social psychology on the ‘audience effect’ –wherein individuals act differently when they are being watched in a more or less conscious attempt to enhance their reputation in the eyes of others, we reasoned that social contexts are indeed likely to produce an increase in performance in typically developing (TD) individuals but that children with ASD would be less susceptible to such audience effects. More specifically, we were interested in testing the idea that susceptibility to the audience effect might explain part of the performance gap between children with and without autism in Theory of Mind (ToM) tasks, which are typically administered by a human experimenter. We tested this hypothesis by comparing performance on a ToM task administered in a social vs. a non-social setting. We found that ASDs and controls performed similarly when the task was administered using a non-social medium. However, control participants outperformed children with autism when an experimenter administered the task. Thus, TD controls demonstrated a relative improvement in performance when in the presence of an experimenter that children with ASD did not. The implications of this diminished ‘audience effect’ in ASD are discussed.