Objective: The aim of this study was to compare associative learning in healthy subjects and patients with degenerative cerebellar disease using classical conditioning techniques in a precision grip task ("pick-the-raspberry" task). Methods: In the pick-the-raspberry task subjects gripped a specialised grip rod and had to increase pull force in a linear fashion while the rod was blocked (pull-phase). The pull-phase lasted for a random time interval of 1–5 s until the rod was released (US), comparable to the time at which a raspberry had just been pulled off a bush. Immediately after US, subjects had to antagonise grip behaviour and to diminish pull force rapidly. Biomechanical (grip and load forces) and electrophysiological (m. thenar, m. interossius1/FDI, m. extensor digitorum communis/EDC, m. flexor carpi ulnaris/FCU) data were collected and analysed. An auditory signal (1 kHz, 450 ms) was employed as the conditioning stimulus (CS) 450 ms before US. Using a linear, 2-spline model pull-force data in the CS-US window were fitted and conditioned responses were assessed when the slope of the second spline fell more than 1.5 N/s below the first. Results: A group of young, healthy subjects (N=10, 23 years ± 3) showed in most of the CS-US trials (approx. 2/3) conditioned responses in the pull force. The typical conditioned responses were characterized by a decrease of the activity of the thenar, FDI, EDC and/or an increase of FCU activity, although inter-individual differences within the main pattern were observed. In contrast, patients with cerebellar disease (N=10; 50 years ± 13) showed a reduced number of conditioned responses (approx. 1/3) with their main muscle activity pattern remaining unchanged. Conclusions: These results show that grip forces in healthy subjects can be classically conditioned. Patients, however, showed clear difficulties in associating grip behaviour with a preceding additional signal. Supported by Else-Kröner-Fresenius Stiftung (A12/07).