The models of stuttering and delayed auditory feedback (DAF) discussed in this paper are derived from a model of the fluent production of speech in which a rhythmic structure prespecifies the intervals between vowels of stressed syllables. The same rhythmic structure enables a prediction to be made concerning when vowels of stressed syllables will be auditorily perceived. In both stuttering and speech produced under DAF conditions, it is suggested that an incorrect prediction is made concerning when the vowel will be perceived relative to its production. The proposed model of stuttering can account for a variety of other phenomena. Finally, a preliminary model of the onset of stuttering in children is proposed.