The implementation of global REDD+ programs could potentially be in the form of a payment for environmental service (PES) scheme of historic dimensions. However, efficiency is a central issue, and knowledge of patterns of opportunity costs among rural populations likely to be affected by REDD+ remain scant. We applied an auction and standard contingent valuation instrument to assess the WTA compensation among a split sample of rural households in the Brazilian Amazon for stopping deforestation activities. The results are compelling and showed that the auction framing format effectively reduced means and variances of stated WTA. Mean WTA is reduced by almost a factor of two, and the sample variance is reduced almost fourfold. The auction framing not only effectively reduces the proportion of very high stated WTA observations, but generally reduces upward biases across the distribution. We compared the elicited measures to standard foregone income measures of opportunity costs. While foregone agricultural income did imply a higher WTA, other determinants affected different groups. Amongst poorer households, REDD+-based transfers could be an attractive low-risk income source, and due to limited alternatives, stated WTA is relatively close to foregone income measures. For higher income groups, larger gap between WTA and opportunity cost exists, likely due to the existence of alternative options for labour and inputs, which reduces the true opportunity costs of stopping deforestation. This study is the first of its kind to apply such a framing in a WTA study, and the robustness of results is remarkable and promising given the developing country context in which it was implemented.