Reducing carbon emissions through avoided deforestation and forest degradation and enhancement of carbon stocks (REDD+) is key to mitigating global climate change. The aim of REDD+ social safeguards is to ensure that REDD+ does not harm, and actually benefits, local people. To be eligible for results-based compensation through REDD+, countries should develop national-level safeguard information systems to monitor and report on the impacts of REDD+. Although safeguards represent a key step for promoting social responsibility in REDD+, they are challenging to operationalize and monitor. We analyzed the impacts of different types of REDD+ interventions (incentives vs. disincentives) on key safeguard-relevant indicators, i.e., tenure security, participation, and subjective well-being, as well as on reported forest clearing. We used household-level data collected in Brazil, Peru, Cameroon, Tanzania, Indonesia, and Vietnam from approximately 4000 households in 130 villages at two points in time (2010-2012 and 2013-2014). Our findings highlight a decrease in perceived tenure security and overall perceived well-being over time for households exposed to disincentives alone, with the addition of incentives helping to alleviate negative effects on well-being. In Brazil, although disincentives were associated with reduced reported forest clearing by smallholders, they were the intervention that most negatively affected perceived well-being, highlighting a clear trade-off between carbon and noncarbon benefits. Globally, although households exposed to REDD+ interventions were generally aware of local REDD+ initiatives, meaningful participation in initiative design and implementation lagged behind. Our analysis contributes to a relatively small literature that seeks to operationalize REDD+ social safeguards empirically and to evaluate the impacts of REDD+ interventions on local people and forests.