The study addresses the issue of the role of the cerebellum in human withdrawal-reflex conditioning by comparing data from patients with pure cerebellar diseases (CBL, n = 10) and from cerebellar patients showing additional extracerebellar symptoms (CBL+, n = 10) with those from 11 control subjects (CTRL). During recording sessions, the standard delay-conditioning paradigm with paired-trials was used with tone as the conditioned stimulus (CS). Parameters of the conditioned muscle responses are analyzed in an accompanying paper. Here, we focus on the unconditioned muscle response. A train of current pulses (unconditioned stimulus, US) evoked a lower-limb withdrawal reflex (unconditioned response, UR), which was recorded electromyographically from leg muscles. During the recording sessions with CTRL subjects, UR amplitudes decayed from initially 100% to approximately 50% at the end of the session. This type of decay was clearly less pronounced in the CBL group and minimal in the CBL+ group. Furthermore, the CBL group exhibited UR onsets that were delayed by 20 ms compared with those from CTRL subjects. Although the ranges of measurements characterizing the URs of a given cerebellar patient tested in the paired-trial paradigm overlapped with those of control subjects, the statistically significant differences observed at the group level suggest deficits in the performance of the reflex responses. The delayed URs in patients and the different type of decay of UR amplitudes in repetitively evoked withdrawal reflexes constitute evidence that the cerebellum is critically involved in the control of these UR parameters.