Questions: We examined the drivers of tree species variation across Neotropical non-flooded evergreen forests (NNFEFs) to answer the following questions: can floristic groups be differentiated based on environmental predictors? How do bioclimatic, topographic, edaphic predictors and dispersal barriers contribute to explain the floristic variation throughout NNFEFs?. Location: Neotropical region. Methods: Based on 1,843 sites (circular areas with a diameter of 10 km), 15,072 species and 509,793 occurrence records of trees, as well as on environmental variables (42 bioclimatic, 13 edaphic and four topographic variables) and dispersal barriers (based on ecological and geographical dispersal suitability), we tested whether environmental predictive variables can discriminate NNFEF floristic groups, and built canonical models and variation partitioning to assess which variables contributed most to the floristic variation. Results: Despite extensive overlap in predictive variables, Amazon and Atlantic Forest were the most differentiated among the nine NNFEF groups. Floristic variation along NNFEFs was mainly determined by environmental factors (54.1%), with topographic and edaphic variables, mainly topographic wetness index and pH respectively, representing the most important predictors followed by a combination of environmental factors and dispersal barriers (22.7%). The fraction of pure dispersal barriers also contributed significantly to our model (3.7%), especially considering the second canonical axis. Conclusions: The high importance of soil and topographic variables indicates that the species have a relatively narrow niche driven by such factors, suggesting that conservation strategies should not be generalized for NNFEFs. In addition, dispersal barriers do not seem to have prevented floristic exchanges between most NNFEF groups, except in the Atlantic Forest.