In this work we present a semi-automated procedure for monitoring deforestation and forest degradation in the Brazilian Amazon using a multi-temporal dataset of Sentinel-2 sensor. Forest cover degradation in the Brazilian Amazon region is mainly due to selective logging of intact/un-managed forests and to wildfires. The study area covers part of a Sentinel-2 sensor scene located in the State of Mato Grosso, in the “deforestation arc” of the Brazilian Legal Amazon. We selected three cloud-free Sentinel-2 images acquired on 21st June, 1st August and 10th September 2016. We generated soil, vegetation and shade fraction images for highlighting the deforested, burned and selectively logged areas. Our analysis shows that deforestation and forest degradation by fire can be mapped using object based analysis. On the other hand, forest degradation by selective logging can be mapped using a pixel based classification of fraction images. Our results allowed the estimative of recent deforestation processes in old growth forests: 1,000 ha between 21st June and 1st August and 900 ha between 1st August and 10th September 2016. The burned forest areas corresponded to 10,700 ha between 21st June and 1st August and to 22,800 ha between 1st August and 10th September 2016. Degraded forest areas due to selective logging added to 135,000 ha as mapped in the image dated 1st August 2016 with 17,300 ha of new areas mapped in the image dated 10th September 2016. The proposed approach shows great potential for monitoring deforestation and forest degradation activities by selective logging and fires using the Sentinel-2 multi-temporal dataset, facilitating the implementation of actions of forest protection in Amazon region.