The aim of the present study was to investigate the speech preparation processes of adults who stutter (AWS). Fifteen AWS and fifteen adults with fluent speech (AFS) participated in the experiment. The event-related potentials (ERPs) were recorded in a foreperiod paradigm. The warning signal (S1) was a color square, and the following imperative stimulus (S2) was either a white square (the Go signal that required participants to name the color of S1) or a white dot (the NoGo signal that prevents participants from speaking). Three differences were found between AWS and AFS. First, the mean amplitude of the ERP component parietal positivity elicited by S1 (S1-P3) was smaller in AWS than in AFS, which implies that AWS may have deficits in investing working memory on phonological programming. Second, the topographic shift from the early phase to the late phase of contingent negative variation occurred earlier for AWS than for AFS, thus suggesting that the motor preparation process is promoted in AWS. Third, the NoGo effect in the ERP component parietal positivity elicited by S2 (S2-P3) was larger for AFS than for AWS, indicating that AWS have difficulties in inhibiting a planned speech response. These results provide a full picture of the speech preparation and response inhibition processes of AWS. The relationship among these three findings is discussed. However, as stuttering was not manipulated in this study, it is still unclear whether the effects are the causes or the results of stuttering. Further studies are suggested to explore the relationship between stuttering and the effects found in the present study.