Tropical forests play a central role in reaching the climate change mitigation goals of the Paris agreement. Over the last decade, hundreds of subnational initiatives that aim to reduce deforestation related emissions (REDD+) have been implemented across the tropics which need to be evaluated and monitored. This makes it important to develop and compare different approaches to quantifying impacts on carbon emissions and associated socio-economic trade-offs and synergies. This thesis compared methods that assess deforestation before and after the start of REDD+, in- and outside REDD+ initiatives, and across different scales. These assessments require (spatial) data on forest area and changes therein, but these data contain uncertainties which influence the REDD+ evaluation results. These uncertainties need to be accounted for in REDD+ assessments. The causes of deforestation differ across and within the different study sites, thus calling for locally tailored approaches when designing REDD+ interventions. Apart from the outcomes of these interventions on forest change, well-being outcomes need to be addressed and assessed too in order not to harm forest-dependent livelihoods.