This paper reports a study of the ability to reproduce stress in a nonsense syllable imitation task by adolescent speakers with autism spectrum disorders (ASD), as compared to typically developing (TD) age-mates. Results are reported for both raters' judgments of the subjects' stress production, as well as acoustic measures of pitch range and duration during stressed and unstressed syllable production. Results reveal small but significant differences between speakers with ASD and typical speakers in both perceptual ratings of stress and instrumental measures of duration of syllables. The implications of these findings for understanding prosodic deficits in ASD are discussed.