Depending on the strength of environmental filtering and competitive exclusion, successful colonizers of plant communities show varying degrees of similarity to resident species with respect to functional traits. For the present study, colonizer's performance was assessed in relation to the degree of fit with the resident community, and in addition, in relation to the community's trait profile and the environmental factors at the study locations. The two-year field experiment investigated the relative growth rates of 130 species that had been transplanted into German grassland communities varying in intensities of land-use. The transplanted species were selected in accordance with the following scenarios: species with highly similar or dissimilar traits to residents, species with highest degree of co-occurrence with resident species and species chosen randomly from the local species pool. The performance of transplanted phytometers depended on the scenario according to which the species were selected, on community trait diversity, and in addition, often on the interaction of both and on land use intensity. The total amount of explained variance in performance was low, but increased considerably when species identity was taken into account. In general, individuals in the co-occurrence scenario performed better than those selected based on trait information or those selected randomly. Different predictors were important in different seasons, demonstrating a limited temporal validity of performance models.