We study the Dalitz plot of the decay D(+)-->K(-)pi(+)pi(+) with a sample of 15090 events from Fermilab experiment E791. Modeling the decay amplitude as the coherent sum of known Kpi resonances and a uniform nonresonant term, we do not obtain an acceptable fit. If we allow the mass and width of the K(*)(0)(1430) to float, we obtain values consistent with those from PDG but the chi(2) per degree of freedom of the fit is still unsatisfactory. A good fit is found when we allow for the presence of an additional scalar resonance, with mass 797+/-19+/-43 MeV/c(2) and width 410+/-43+/-87 MeV/c(2). The mass and width of the K(*)(0)(1430) become 1459+/-7+/-5 MeV/c(2) and 175+/-12+/-12 MeV/c(2), respectively. Our results provide new information on the scalar sector in hadron spectroscopy.