Monitoring our actions is a key function of the brain for adaptive and successful behavior. Actions can be discrete such as when pressing a button, or continuous, such as driving a car. Moreover, we evaluate our actions as correct or erroneous (performance monitoring) and this appraisal of performance comes with various levels of confidence (metacognition). However, studies of performance monitoring have focused on discrete actions and are mostly agnostic to metacognitive judgments. The objective of this thesis was to extend the study of performance monitoring to more ecological conditions, in which monitoring occurs during continuous motor tasks under various degrees of error and confidence level. We first investigated the role of actions in performance monitoring together with metacognitive judgments, using simultaneous EEG and fMRI recordings. To dissociate the role of motor actions, we designed an experimental paradigm in which subjects had to rate their confidence level about an action that they had either performed themselves (a button press) based on a decision or passively observed (a virtual hand displayed). We found correlates of confidence in both condition, in the EEG and in the supplementary motor area (SMA). Furthermore, we found that subject showed better metacognitive performances when they were the agents of the action. This difference was further emphasized for subjects that showed higher activations of a network previously linked to motor inhibition and comprising the pre-SMA and inferior frontal gyrus (IFG). Our results imply that the SMA plays a primary role in the monitoring of performance, irrespectively of a commitment to a decision and the resulting action. Our findings also suggest that the additional neural processes leading to decisions and actions can inform the metacognitive judgments. In the following chapters, we ask whether electrophysiological correlates of performance monitoring can be found in less experimentally constrained paradigms for which motor output continuous unfolds and visual feedback is not segregated into discrete events. By decomposing the unfolding hand kinematics during a visuo-motor tracking task into periodic acceleration pulses –henceforth referred to as sub-movements, we found three electrophysiological markers that could possibly be linked to performance monitoring. Firstly, we found an ERP in the SMA, time-locked to sub-movements which encoded the deviation of the hand, 110 ms before. Secondly, we found high-gamma activity in the ACC and SMA of epileptic patients, that was phase-locked to sub-movements. Thirdly, we found a transient modulation of mu oscillations over the ipsilateral sensorimotor cortices that depended on sub-movement amplitude. Altogether, these results provide a strong contribution in the understanding of the neurophysiological processes underlying performance monitoring. Our work proposes a methodological framework to study electrophysiological correlates of performance monitoring in less controlled paradigms during which continuous visual feedback has to be constantly integrated into motor corrections. In the conclusion chapter, we propose a way of extending current models of performance monitoring and decision making to explain the findings of this thesis by considering continuous motor tasks as a succession of decision making processes under time pressure and uncertainty.