A basic tenet of the Marr-Albus theory of motor learning is that the learning process involves concurrent activation of the climbing fibre and mossy fibre inputs to the cerebellum. This theory was tested by administering harmaline, a drug which causes climbing fibres to fire at their maximal rate of 8-12 Hz, to rabbits during a classical conditioning protocol. Harmaline significantly reduced the proportion of conditioned nictitating membrane responses on each of 4 successive training days. When harmaline was given to already-conditioned rabbits, the proportion of conditioned responses fell sharply to the day 1 level: however, this was still a higher proportion of conditioned trials than that which was seen with naive animals trained for 4 days with harmaline. It is therefore concluded that harmaline disrupts the acquisition of classically-conditioned nictitating membrane responses but does not prevent the expression of an already learnt response. These findings are therefore consistent with the proposal in the Marr-Albus theory that the climbing fibres play an essential part in motor learning.