Background/Question/Methods As the concept of payment for ecosystem services (PES) has grown, so too has the need to develop a stronger scientific basis for these programs. This includes determining which land uses and management strategies will maximize production of the desired ecosystem services and understanding the potential trade-offs among the services produced under each land use. Results from my research suggest that these answers may not always be straightforward; outcomes may not conform with long-standing notions about desirable land covers and uses, and they may exhibit strong regional and local variability that require ecosystem services policies to be developed at those scales. I draw on examples ranging from global to local levels to illustrate these issues. Results/Conclusions Globally, there has been a push towards increasing the amount land area under forest cover, with the assumption that this transition will have a range of positive biophysical outcomes that will enhance the production of ecosystem services. However, in the case of tree plantations established in grassland and shrubland ecosystems, research on carbon-water tradeoffs has suggested that this transition can be effective for carbon sequestration and storage, but comes at the cost of a loss of streamflow which can compromise watershed sustainability. At the regional level, these trade-offs can be strongly influenced by the characteristics of the ecosystems in which PES programs are developed. In the case of Ecuador, high elevation grasslands that are characterized by soils with exceptionally large carbon contents and water-holding capacities have been the site of both carbon- and water-focused PES programs. Land use and management have shifted in response, leading to an increase in the establishment of tree plantations intended to generate carbon credits and the alteration of traditional burning and grazing regimes with the intent of enhancing the generation of water resources. However, the biophysical outcomes of these and other changes in land use and management are poorly understood, leaving the potential for well-intended land-use changes to produce undesired consequences. These cases illustrate the need to base payment for ecosystem services programs in sound science and to approach choices about land use and management with empirical grounding rather than assumptions about the ecosystem services they might produce. Ultimately, the success of payment for ecosystem services programs relies not only on connecting ecosystem suppliers and consumers, but also on connecting particular land use systems with the desired set of biophysical outcomes.