Ecological functions, including pollination services, can be successfully restored in degraded ecosystems. In this study we propose a procedure for selection of target species to be employed in the restoration of plant-pollinator interactions in agricultural landscapes, which is illustrated by a case study located in the Tandilia System, in the Southern Pampa grasslands of Argentina. Based on information from a large pollination network dataset of 12 hills, composed of 172 pollinators and 96 plants (metaweb), we identified the plant species playing a major role in the maintenance of pollination mutualisms. We obtained a ranking of interaction frequency for each of the 96 plant species of the metaweb, and selected native plants that received more than 100 individual flower visitors. The targeted species were evaluated using ten criteria related to ecological, technical and cultural characteristics, and then ranked according to their suitability for use in restoration projects (species rank, SR). From the total number of plant species (96) registered for all sampled hills, we identified a total of 24 plant species that represented 90% of the total interactions and 25% of the plant species recorded in the metaweb. Six of the 24 selected species were excluded since they were non-native, leaving 18 targeted native plant species. The SR value ranged between 2.68 and 8, with a mean of 4.63 +/- 1.40. Two genera presented the highest SR values (Eryngium sp. and Baccharis sp.), and are recognized as potential candidates for restoration of other ecosystems. A rank that evaluates plant species capable of rehabilitating plant-pollinator interactions will form a solid basis for planning restoration projects, which are crucial for biodiversity rehabilitation and conservation.