Although acclaimed as a biofuel crop with high potential to sustainably replace fossil fuels, Jatropha curcas L. remains a poorly studied plant. Reliable yield assessments with conventional methods require agroclimatic and physiological knowledge, which is not yet available for Jatropha. To fill this gap, we tested a novel two-step approach integrating knowledge from biogeography and population biology with available Jatropha field data. In the first step, using MaxEnt, a widely implemented model in biogeography, we predicted Jatropha fitness in response to climate by relating natural occurrence recorded in herbaria with bioclimatic geodatasets. In the second step, we relied on population biology principles supported by seed mass addition experiments to relate fitness to reproductive potential, hence seed yield. Jatropha seed yield in response to climate was mapped worldwide for actual (1950–2000 average) and future (2020) climate conditions. The modelled Jatropha seed yield was validated against a set of on-field yield assessments (R2=0.67, P<0.001). The discrepancies between estimated and measured yields were partially explained by model uncertainties, as quantified by the sensitivity analysis of our modelling (R2=0.57, P=0.001). Jatropha has a pan-tropical distribution, plus specific adaptability to hot temperate areas. Climate variables most significantly affecting modelled yield response were annual average temperature, minimum temperature, annual precipitation and precipitation seasonality.