Background Patients with cervical dystonia (CD) present with an impaired performance of voluntary neck movements, which are usually slow and limited. We hypothesized that such abnormality could involve defective preparation for task execution. Therefore, we examined motor preparation in CD patients using the StartReact method. In this test, a startling auditory stimulus (SAS) is delivered unexpectedly at the time of the imperative signal (IS) in a reaction time task to cause a faster execution of the prepared motor programme. We expected that CD patients would show an abnormal StartReact phenomenon. Methods Fifteen CD patients and 15 age matched control subjects (CS) were asked to perform a rotational movement (RM) to either side as quick as possible immediately after IS perception (a low intensity electrical stimulus to the II finger). In randomly interspersed test trials (25%) a 130 dB SAS was delivered simultaneously with the IS. We recorded RMs in the horizontal plane with a high speed video camera (2.38 ms per frame) in synchronization with the IS. The RM kinematic-parameters (latency, velocity, duration and amplitude) were analyzed using video-editing software and screen protractor. Patients were asked to rate the difficulty of their RMs in a numerical rating scale. Results In control trials, CD patients executed slower RMs (repeated measures ANOVA, p<0.10−5), and reached a smaller final head position angle relative to the midline (p<0.05), than CS. In test trials, SAS improved all RMs in both groups (p<0.10−14). In addition, patients were more likely to reach beyond their baseline RM than CS (χ2, p<0.001) and rated their performance better than in control trials (t-test, p<0.01). Conclusion We found improvement of kinematic parameters and subjective perception of motor performance in CD patients with StartReact testing. Our results suggest that CD patients reach an adequate level of motor preparation before task execution.