Singing as a fluency-enhancing mechanism is well-established. The fluency derived by singing has been attributed to a reduced speech rate, memorized material, semantically reduced content, and an imposed rhythm. In this study, we attempted to address each of these explanations. 12 participants who stuttered were instructed to read or sing each of four different passages under the following conditions: reading at a normal rate, reading at a fast rate, singing at a normal rare, and singing at a fast rate. Participants exhibited a statistically significant increase in disfluencies while reading, i.e., participants displayed a 75% reduction in disfluency in the singing condition relative to the reading condition. There was no difference in stuttering frequency with rate conditions. Current findings suggest that stutterers are capable of internally generating fluent speech production by imposing idiosyncratic melodic structures or some derivation of melody when asked simply to sing. There is no claim that these participants were singing, as skills and capabilities varied tremendously, only that participants achieved dramatic enhancement of fluency after they were just asked to sing. Thus, the only intervening variable was the instruction to sing, which suggests the attempt to follow the instruction, no matter how futile, generated fluent speech. Since fluency was maintained in both the normal and fast rates of production, alternate central mechanisms must be held accountable for these findings.