Motor learning disturbances have been shown in diseases involving dopamine insufficiency such as Parkinson's disease and schizophrenic patients under antipsychotic drug treatment. In non-human primates, motor learning deficits have also been observed following systemic administration of raclopride, a selective D2-receptor antagonist. These deficits were characterized by persistent fluctuations of performance from trial to trial, and were described as difficulties in consolidating movements following a learning period. Moreover, it has been suggested that these raclopride-induced fluctuations can result from impediments in grouping separate movements into one fluent sequence. In the present study, we explore the hypothesis that such fluctuations during movement consolidation can be prevented through the use of sumanirole - a highly selective D2 agonist - if administered before raclopride. Two monkeys were trained to execute a well known sequence of movements, which was later recalled under three pharmacological conditions: (1) no drug, (2) raclopride, and (3) sumanirole+raclopride. The same three pharmacological conditions were repeated with the two monkeys, trained this time to learn new sequences of movements. Results show that raclopride has no deleterious effect on the well known sequence, nor the sumanirole+raclopride co-administration. However, results on the new sequence to be learned revealed continuous fluctuations of performances in the raclopride condition, but not in the sumanirole+raclopride condition. These fluctuations occurred concurrently with a difficulty in merging separate movement components, known as a "chunking deficit". D2 receptors seem therefore to be involved in the consolidation of new motor skills, and this might involve the chunking of separate movements into integrated motor sequences.