Systematic recordings were made at regular intervals (6, 9, 12, and 15 months) and utterances were transcribed and analyzed from speech samples of children at risk and not at risk to stutter. Children were audio and video recorded interacting with their own mother for 10‐min play periods. The purpose of the study was to gather quantitative information on fluency development prior to the diagnosis of a stuttering disorder to determine whether or not children can be differentiated based on vocal development. Wideband spectrograms and formant tracking displays were used to analyze oral development (plotting first and second formant frequencies of the steady‐state part of the vowel) and the other measurement pertained to laryngeal‐oral coordination (analyzing voice onset time in stop consonants). Acoustic analyses were performed using the Kay Elemetrics CSL workstation. Control subjects had more elaborated vowel systems than children who end up stuttering in their vocalizations. Pre‐stuttering children also displayed repetitive disfluencies in their vocal play. [Work supported by NIH and ASHF.]