We examined whether resting-state and task-related oscillations differently predict the practice effect during virtual tool-use training in young (YA) and older (OA) adults. Thirty-seven YA (Mage: 23.64, SD: 7.07) and forty-one OA (Mage: 68.92, SD: 4.49) learned to control a virtual gripper to grasp a virtual object. The training was organized in two blocks (120 trials each). In one block, vibrotactile feedback was applied to the right hand's palm, thumb, and index fingers through a CyberTouch II glove when the tool touched the object. In the other block, participants only received visual feedback. The total practice effect was calculated per block as the relative gripper size at the end of the block compared to the starting gripper size. The latter has been changed adaptively by decreasing the width of the tool in a 3 down / 1 up staircase procedure. Resting-state EEG was collected before the first and second training block (pre-test) and task-related EEG was collected during first and second training blocks. Results revealed that higher relative beta power at rest predicts stronger practice effect in YA, while there is an inverse association between higher relative beta power at rest and lower practice effect in OA. In addition, higher task-related relative theta power in frontal and occipital regions and lower task-related relative alpha power in frontal and parietal regions predicted stronger practice effect in YA. We, therefore, conclude that the stronger practice effect during virtual tool-use training can be predicted by resting-state relative beta power, higher task-related relative theta power and lower task-related alpha across different regions in YA but not in OA.