The research literature on stuttering is extensive, spanning etiology, natural history, phenomenology, and treatment. In this review, existing empirical knowledge is distilled by attending only to replicated findings. These are designated as "facts." Facts concerning the natural history of idiopathic stuttering highlight early childhood onset, probability of recovery, and importance of a positive family history of stuttering. Corroborated evidence on stutterer-nonstutterer differences concerns intelligence distribution, speech development, central auditory function, and sensory-motor response. Predictable changes in stuttering frequency, and even total elimination of stuttering, occur under a remarkable variety of conditions. Review of therapies revealed two that satisfied the most stringent criteria for good treatment. Various theoretical positions are examined for their fit with the established facts. A model of stuttering as a genetically determined reduction in central capacity for efficient sensory-motor integration is preferred, provided acquisition of secondary symptoms is attributed to instrumental learning.