The institutional predecessor of REDD+ is the integrated conservation and development project (ICDP) that combines restrictions on forest access and conversion (negative interventions) with non-conditional direct benefits (positive interventions) to compensate local stakeholders for income losses from those restrictions. The idea of REDD+ was to improve on the ICDP model with a different kind of positive intervention: conditional direct benefits, often known as payments for environmental services or PES. How has this idea played out in reality? In a sample of 17 (out of 377) active REDD+ initiatives across the global South, we identified the combinations of interventions actually deployed and elicited household assessments of how those interventions affected their land use decisions with respect to forests. We found that 71 % of the households in our sample had participated in some number of forest interventions ranging from one to ten. About a quarter of those households were offered conditional direct benefits, most often in combination with non-conditional direct benefits. Nearly half of the households had received only non-conditional direct benefits. Many of those households were also subject to restrictions of various kinds. Thus, rather than abandoning the well-established ICDP approach in favor of the conditional incentives that conceptually define REDD+, most initiative proponents opted to deploy multiple interventions. Their approach is validated by our finding that the likelihood a household reports that the interventions caused them to adopt land use changes that could be classified as reducing carbon emissions is positively and significantly related to the number of interventions that they experienced, but not affected by whether any of those interventions are conditional. We also find that restrictions play an important role: 37 % of the households were subject to at least one negative intervention, and those households were significantly more likely t