A unique pattern of weed exclusion was found under the canopy ofDelonix regia, which was planted in many places as an ornamental tree in the south of Taiwan. A quadrat method was employed to examine the botanical composition between the area underneathD. regia and its adjacent control grassland. The number of species and coverage of understory species were significantly lower in the area of theD. regia than that of the grassland, indicating the growth of understory species was suppressed byD. regia. A series of aqueous extracts of leaves, flowers, and twigs ofD. regia were bioassayed against three species to determine their phytotoxicity, and the results showed highest inhibition in the flowers. A water-culture experiment indicated the aqueous extract of flowers ofD. regia on two local understory species (Isachne nipponensis andCentella asiatica) inhibited growth of both species by more than 70%. The phytotoxicities of fallen leaves and flowers ofD. regia were not significantly affected by temperature. When the plant material was subjected to temperatures above 70°C, however, phytotoxicity was decreased, indicating that the allelopathic nature ofD. regia could easily be decomposed by fire. By means of paper, thin-layer, and high-performance liquid chromatography, and UV-visible spectrophotometry, responsible phytotoxins present in leaves, flowers, and twigs ofD. regia were identified as 4-hydroxybenzoic, chlorogenic, 3,4-dihydroxybenzoic, gallic, 3,4-dihydroxycinnamic, 3,5-dinitrobenzoic, and L-azetidine-2-carboxylic acids, and 3,4-dihydroxybenzaldehyde. The findings of bioassays and the number and amount of responsible allelopathic compounds found inD. regia are well correlated, thus permitting the conclusion that the exclusion of understory plants under the canopy ofD. regia trees was due primarily to the allelopathic effect of the fallen flower, leaves, and twigs of theD. regia. A possible mechanism of action is discussed.