Compensating forest users for the opportunity costs of foregoing deforestation and degradation was one of the original distinguishing features of REDD+ (Reducing Emissions from Deforestation and Degradation). In the early days of REDD+, such costs for tropical smallholders were believed to be quite low, but this has increasingly been questioned. A decade after the concept was proposed, direct payments to forest stakeholders remain rare, while concerns about safeguarding livelihoods are increasing. Households facing restrictions on forest-based activities will have to be compensated, yet evidence on actual costs to households, their distribution, and implications for efficiency and equity is limited. We estimate smallholder opportunity costs of REDD+ in 17 sites in six countries across the tropics. We use household data collected from multiple sites in multiple countries using a uniform methodology. We find that opportunity costs per tCO2 emissions from deforestation are less than the social costs of tCO2 emissions ($36) in 16 of the 17 sites; in only six of the sites, however, are opportunity costs lower than the 2015 voluntary market price for tCO2 ($3.30). While opportunity costs per tCO2 are of interest from an efficiency perspective, it is opportunity costs per household that are relevant for safeguarding local peoples' income. We calculate opportunity costs per household and examine how these costs differ for households of different income groups within each site. We find that poorer households face lower opportunity costs from deforestation and forest degradation in all sites. In a system of direct conditional payments with no transactions costs to households, poorer households would earn the highest rents from a system of flat payments. Our findings highlight that heterogeneity and asymmetrical distribution of opportunity costs within and between communities bear important consequences on both equity and efficiency of REDD+ initiatives.