Abstract It has been increasingly argued that ecological restoration should focus more on targeting ecosystem services than on species composition of reference ecosystems. In this sense, the role that species play on community assembly and functioning through their functional traits is very relevant, because effect traits mediate ecosystem processes, ultimately resulting in provision of ecosystem services. One major challenge in ecological restoration is to know which species to use that will deliver the target ecosystem services. We developed an algorithm to select the minimum set of species that maximize the functional richness (FRic) and the functional redundancy (FR) of the restored community, a proxy for the provision of multiple ecosystem services and the resilience of the system to environmental changes, respectively. For this, we simulated the restoration of 24 riparian woody communities of the Brazilian Cerrado. Using the species pool of each original local community, we ran restoration simulations for gradually increasing species richness until reaching the total species richness of the original local community. We computed FRic and FR for each simulated restoration community using the traits specific leaf area, maximum plant height and seed mass. Our simulation results indicate that multiple ecosystem services could be restored with an average of 66% of the species of the original community. Moreover, an average of 59% of the species would be needed to restore communities resilient to environmental changes. Our approach contributes to solving one of the major challenges of ecological restoration, which is defining how many and which species should be used to achieve functional targets. We believe this approach can help in projects of restoration by enabling restoration practitioners to select minimum alternative sets of species that optimize the provision of multiple ecosystem services in a resilient restored ecosystem.