Landholders are generally assumed to be willing to participate in payments for ecosystem service (PES) schemes if the offered payment exceeds the opportunity cost of participation. The calculation of opportunity costs is often based on historic financial data such as net returns of the formerly practiced land use. Reliable estimates of opportunity costs are required especially in flexible, cost-aligned payment schemes with differentiated payments at the farm scale. We question whether opportunity cost estimates that do not consider personal landholder characteristics such as risk considerations, information access and non-monetary personal preferences (e.g. for traditional land use practices) are sufficient to explain a landholder's decision to enrol land in PES. To test these assumptions, a PES adoption model was developed for hypothetical adoption decisions by 178 landholders in Costa Rica. The model explained up to 73.5% (Nagelkerkes pseudo R2) of adoption variance. The results confirm that adoption is not determined by financial costs alone. Trust in state institutions, for example, was highly significant. The results call for more integrated methods of opportunity cost estimation such as inverse auctions. Their strength lies, among others, in that all adoption determinants are potentially expressed in the landholder's bid.