Urban areas serve as hubs for non-native plant introductions, but the extent to which these introductions have homogenized city floras globally remains unexplored. We analyzed species inventories from 553 cities across six continents and show that non-native plants increase mean pairwise floristic similarity by ~50%, from a Jaccard index of 0.11 for native species alone to 0.16 when non-natives are included. Non-natives increase intracontinental similarity by 25%, but boost intercontinental similarity by 423%, due to intercontinental native floras sharing few species. City connectivity (indexed by air-travel frequency) amplifies non-native similarity more than native similarity, while geographic distance constrains native similarity more strongly, consistent with dispersal limitation. Contrary to our original hypothesis, climatic distance constrains non-native similarity as strongly as native similarity, confirming that climate is a hard environmental filter regardless of species’ origin. Meanwhile, urban socio-economic and physical features also affect native and non-native floristic similarity comparably. Regional analyses reveal pronounced asymmetries, with non-native species strongly homogenizing cities in Europe, North America, and Oceania, whereas Asian cities show weak intercontinental convergence. Decomposing similarity into ecological components reveals distinct mechanisms: homogenization between city pairs that involve European cities is dominated by species transfers (taxa native in one region but naturalized in another), reflecting colonial and trade legacies, whereas North American-Oceanian convergence is driven by a shared pool of widespread non-native species. These findings demonstrate that human-mediated plant movement has become a dominant force reshaping global urban flora, with effects modulated by connectivity, environmental filtering, and region-specific historical legacies.