Recent studies revealed an impairment of patients with cerebellar lesions in cognitive associative learning paradigms, which could not be explained by motor deficits or increased attentional demands during performance of the motor part of the particular task. To further test the role of the cerebellum in cognitive associative learning a visuomotor associative learning paradigm was conducted, which showed some similarities to reflex conditioning. Subjects had to learn the association between a visual stimulus (i.e., a colored square) and a correct motor response (i.e., pressing a left or right target button on a specialized keyboard) in a stimulus-response paradigm. The correct side of the response was always indicated by a growing circle. The stimuli were presented in two forms: (1) the colored square ("conditioned stimulus", "CS") preceded the growing circle ("unconditioned stimulus", "US") and coterminated with the "US" (delay-condition) or (2) the offset of the colored square was separated from the onset of the circle by a short time interval (trace-condition). Eight patients with degenerative cerebellar disorders and eight age-, sex-, IQ-, and education level-matched controls participated. Both control subjects and cerebellar patients showed significant effects of learning based on reduction of decision times during performance of the task. The explicit knowledge of the correct motor response associated with each of the presented colors, however, was significantly better in control subjects. Furthermore, within-group comparisons revealed that control subjects performed significantly better in the trace-condition compared to the delay-condition, whereas no significant difference was obtained in the cerebellar group. Part of the findings in controls was explained by an age-related decline in the delay-condition. In addition, elderly control subjects appeared to benefit from cue-effects and timing-effects (i.e., fixed trace interval) in the trace-condition resulting in reduced decision times. Neither cue- nor timing-related effects fully accounted for differences between cerebellar patients and controls. The lack of a better performance in the trace-condition compared to the delay-condition in cerebellar patients suggests deficits in learning the stimulus-response association.