Amygdala, being a major emotion processing brain region, is particularly responsive to faces and, in this context, exhibits abnormal reactivity and connectivity in affective and anxiety disorders. Down-regulation of amygdala as a form of voluntary emotional control can be achieved using real time fMRI neurofeedback (NFB) by implementing different explicit regulation strategies. As per recent evidences, this could be possible when subjects are oblivious about the training and the behavioral changes associated with it. To examine this concept, we applied a novel closed-loop NFB protocol in 64 healthy participants who performed four amygdala closed-loop neurofeedback runs within one scan session. 32 subjects assigned to the explicit group were informed about the NFB protocol and instructed to either reduce the fearful facial expression (fear-down) or increase the happiness of the presented face stimuli (happy-up). 32 subjects assigned to the implicit group were instructed to carefully observe the emotional faces, i.e. without receiving any information about the NFB setup. fMRI measurements were conducted on a 3T Philips Achieva scanner. All the subjects performed four NFB runs (20s baseline + 40s regulation) using OpenNFT, an open source python/matlab based program. Preliminary evidence suggests that explicit and implicit closed-loop amygdala NFB leads to a significant comparable decrease in amygdala activation in response to fearful faces. This finding is relevant as this approach could open up new possibilities of NFB training in clinical conditions where patients have restricted cognitive ability to apply explicit regulation strategies.