Self-contained ear-level devices delivering altered auditory feedback (AAF) for the application with those who stutter have only been recently developed. The paper examines the first therapeutic application of self-contained ear-level devices in three experiments. The effect of the device on the proportion of stuttered syllables and speech naturalness was investigated following initial fitting and at 4 months post-fitting. Three experiments were undertaken: In Experiment 1, the effect of a self-contained in-the-ear device delivering AAF was investigated with those who stutter during reading and monologue. Two adolescents and five adults who stuttered read and produced monologue with and without a device fit monaurally. The device provided a frequency shift of +500 Hz in combination with a delayed auditory feedback of 60 ms. Custom-made ITC and CIC devices were fabricated for four adults and four youths in Experiment 2. The effect of group (i.e. youth versus adult), time (i.e. initial fitting versus 4-month follow-up), speech task (i.e. reading versus monologue), and device (i.e. present versus absent) on stuttering rate was examined. In Experiment 3, 15 naive listeners rated the speech naturalness of speech produced by the participants in Experiment 2. Speech samples from six conditions were rated: reading and monologue without the device at the initial visit, reading and monologue with the device at the initial visit, and reading and monologue with the device at 4 months. In Experiment 1, the proportion of stuttered syllables was significantly (p=0.011) reduced by approximately 90% during reading and 67% during monologue with the device relative to no device. Only a significant main effect of device (p=0.0028) was found in Experiment 2. That is, stuttering rate was significantly reduced with the device in place regardless of speech task or group and remained so 4 months later. In Experiment 3, speech samples generated while wearing the device were judged to be more natural sounding than those without the device (p <0.0001) for reading and monologue with both adults and youths. There was no significant difference between the mean naturalness ratings of speech samples generated during the initial fitting with the device relative to that at 4 months with the device (p >0.05) in all cases except with the youths while engaged in monologue. For that condition, raters judged the speech produced at the initial fitting as more natural. Conclusions: These findings support the notion that a self-contained in-the-ear device delivering AAF assists those who stutter. With the device in place, stuttering is reduced and speech produced is judged to be more natural than with out the device.