Purpose: Speech-language pathologists’ (SLPs) ability to measure stuttering reliably has been of interest over time. However, speech samples used in stuttering measurement research have varied in duration. This study was undertaken to examine whether the duration of speech samples influences the reliability of measurement of severity of stuttering by SLPs. Methods: Ten specialist SLPs rated 27 audio, English speech samples of three different durations (9 × 1-minute, 9 × 3-minutes, 9 × 5-minutes) of adults who stutter, using a 9-point severity rating (SR) scale. Results: The speech sample durations produced similar results when using the SR scale to measure severity of stuttering. Thus, samples of 1, 3 and 5-minute durations were found to be equally appropriate for reliability research and training purposes. Variability was found to be larger in the moderate severity range than the mild and severe ranges. Conclusions: Data trends suggest that SLPs and researchers should focus more attention on practice and training in the middle ranges of the SR scale, due to increased variability in this range.