Watermelon [Citrullus lanatus (Thunb.) Matsum & Nakai] grafting is commonly used for management of diseases caused by soilborne pathogens; however, little research exists describing the effect of grafting on the weed-competitive ability of watermelon. Field experiments determined the response in yield, fruit number, and fruit quality of grafted and nongrafted watermelon exposed to increasing densities of Palmer amaranth (Amaranthus palmeri S. Watson). Grafting treatments included ‘Exclamation’ triploid (seedless) watermelon grafted on two interspecific hybrid squash rootstocks ‘Carnivor’ and ‘Kazako’, with nongrafted Exclamation as the control. Weed treatments included A. palmeri at densities of 1, 2, 3, and 4 A. palmeri plants per watermelon planting hole (0.76-m row) and a weed-free control. Increasing A. palmeri densities caused significant reductions (P < 0.05) in marketable watermelon yield and marketable fruit number. Watermelon yield reduction was described by a rectangular hyperbola model, and 4 A. palmeri plants planting hole ^-1 reduced marketable yield 41%, 38%, and 65% for Exclamation, Carnivor, and Kazako, respectively. Neither grafting treatment nor A. palmeri density had a biologically meaningful effect on soluble solids content or on the incidence of hollow heart in watermelon fruit. Amaranthus palmeri seed and biomass production was similar across weed population densities, but seed number per female A. palmeri decreased according to a two-parameter exponential decay equation. Thus, increasing weed population densities resulted in increased intraspecific competition among A. palmeri plants. While grafting may offer benefits for disease resistance, no benefits regarding weed-competitive ability were observed, and a consistent yield penalty was associated with grafting, even in weed-free treatments.