The goal of this work was to study speech prosody in Autism Spectrum Disorders (ASD). Increasing our understanding of how prosody is diﬀerent in ASD may be important for characterizing its phenotype, helping prosodic remediation, aiding in diagnosis, and providing outcome measures for treatment research. We compared the prosody of children with High-Functioning Autism (HFA) to those with Speciﬁc Language Impairment (SLI) and Typical Development (TD). Our questions included the following: What statistical features of prosody are signiﬁcantly diﬀerent? How does intonation diﬀer qualitatively, for example regarding the shape of the intonation curves? Can naive listeners reliably detect atypical prosody at the utterance level? After explaining the necessary scientiﬁc background, we ﬁrst matched the groups on age and cognitive measures using a novel approach and new algorithms. Subsequently, we compared the prosody of these matched groups based on acoustic-prosodic features from various known and innovative computational techniques, as well as through perceptual ratings of the children’s utterances from naive listeners. My main contribution to knowledge is that high-functioning autistic children without language impairment diﬀered from typical children on various measures of prosody, whereas the autistic children with language impairment did not diﬀer from children with speciﬁc language impairment on any of our measures after controlling for content features. The utterances of children with either HFA or SLI were also perceived as having higher emotional arousal that of those with TD.