This behavioural study concerns the contribution of active retraining to motor recovery after a standard lesion in the motor cortex, and includes an evaluation of various retraining procedures. These problems have not previously been experimentally analysed in man or animal. Rhesus monkeys (27) were initially trained on two motor tasks which consisted of a pulling task, involving the proximal muscles of the upper limb, and a hand-grip task for the distal musculature. Strength of pulling and hand-grip were measured quantitatively. For brevity, only the hand grip data are described. After a plateau of proficient performance was achieved in both hands (usually 6-8 months), the cortical precentral forelimb area was surgically ablated on one side. Each animal was then randomly assigned to one of four experimental groups or to a sham operative control group. The groups differed with respect to the use of the contralateral and/or ipsilateral forelimb(s) in post-operative motor training on the same task. In addition, to evaluate the contribution of spontaneous post-operative recovery independent of retraining, we started to train two groups immediately after surgery; in the other two experimental groups the weak forelimb remained idle for the first four post-operative months. Combined training of the weak and normal limb, which resulted in 85% recovery in the weak limb, did not differ statistically from training the weak limb alone (79% recovery). This suggests that the critical factor in promoting recovery is training of the weak forelimb, presumably by 'activation' of the damaged hemisphere. The role of the ipsilateral (strong) limb appears negligible. When post-operative training in the weak limb was delayed four months, spontaneous recovery noted one week after the start of delayed training was about 50% compared with 9% recovery after one week in the groups retrained immediately after surgery (P less than 0.001). The 'immediate' groups, however, continued to improve over a six-month period to about 82% of their pre-operative performance. The 'delay' groups, by contrast, exhibited only slight further improvement, reaching a plateau of 67% recovery six months after the start of retraining (10 months post-operatively). This difference in recovery between the immediate and delay groups was significant at the 0.05 level. This confirms that active physical retraining facilitates motor recovery, although the mechanism remains obscure. The data also suggest that, to be most effective, the training should begin as soon as possible after the insult to the brain has occurred.