ObjectivesPersistent developmental stuttering is a speech fluency disorder defined by its symptoms, where the underlying neurophysiological causes remain uncertain. This study examined the underlying neurophysiological mechanisms of the speech planning process, using facilitation in the motor cortex during speech preparation as an analogue.Methodstranscranial magnetic stimulation (TMS) pulses induced motor evoked potentials (MEPs), which were recorded from the tongue. Eighteen adults who stutter (AWS) and 17 adults who do not stutter (ANS) completed three experiments, which involved reading a German prefix+verb utterance from a screen. Each experiment involved 120 trials with three distinct levels of speech production: immediate speech, delayed speech without pacing and delayed speech with predefined pacing. TMS was applied shortly before speech onset. Trial MEPs were normalised to average non-speech MEPs. MEP amplitude, MEP facilitation ratio (amplitude: pre-speech offset) and group difference were the outcomes of interest analysed by multiple regression, as well as speech reaction time analysed by correlation.ResultsMEP values were 11·1%-23·4% lower in AWS than ANS (by standardised Beta), across all three experiments. MEP facilitation ratio slopes were also 4·9%-18·3% flatter in AWS than ANS across all three experiments. Reaction times for AWS were only significantly slower than for ANS in immediate speech and predefined pacing experiments. No stuttering was detected during the trials. The group difference in immediate speech was 100% and 101% greater than the other two experiments respectively.DiscussionWhile performance of both ANS and AWS worsens under disturbed speech conditions, greater disturbance conditions affected controls worse than AWS. Future research and therapy in stuttering should focus on non-disturbed speech.