Ecosystem services and economic returns from semi-arid rangelands are threatened by land degradation. Policies to improve ecosystem service delivery often fail to consider uncertainty in economic returns gained through different land uses and management practices. We apply an analytical framework using stochastic simulation to estimate the range of potential monetary outcomes of rangeland ecosystem services under different land uses, including consideration of the uncertainty and variability of model parameters. We assess monetary and nonmonetary dimensions, including those ecosystem services with uncertain and missing information, for communal rangelands, commercial ranches, game farms and Wildlife Management Areas in southern Kgalagadi District, Botswana. Public land uses (communal grazing areas and protected conservation land in Wildlife Management Areas) provide higher economic value than private land uses (commercial ranches and game farms), despite private land uses being more profitable in their returns from meat production. Communal rangelands and protected areas are important for a broader range of ecosystem services (cultural/spiritual services, recreation, firewood, construction material and wild food), which play a key role in sustaining the livelihoods of the largest share of society. The full range of ecosystem services should therefore be considered in economic assessments, while policies targeting sustainable land management should value and support their provision and utilisation. By forecasting the range of plausible ecosystem values of different rangeland land uses in monetary terms, our analysis provides policymakers with a tool to assess outcomes of land use and management decisions and policies. Copyright © 2016 John Wiley and Sons, Ltd.