A portable, low-frequency acoustic system was used to detect termite infestations in urban trees. The likelihood of infestation was rated independently by a computer program and an experienced listener that distinguished insect sounds from background noises. Because soil is a good insulator, termite sounds could be detected easily underneath infested trees, despite the presence of high urban background noise. Termite sounds could be detected also in trunks, but background noise often made it difficult to identify termite signals unambiguously. High likelihoods of termite infestation were predicted at four live oak (Quercus virginiana Mill, Fagacae), two loblolly pine (Pinus taeda L., Pinacae), and two baldcypress (Taxodium distichum Rich. Pinacae) trees that wood-baited traps had identified as infested with Coptotermes formosanus Shiraki. Infestations also were predicted at two pine trees with confirmed recoveries of Reticulitermes flavipes (Kollar). Low likelihoods of infestation were predicted in four oak trees where no termites were found. Additional tests were conducted in anechoic environments to determine the range of acoustic detectability and the feasibility of acoustically estimating termite population levels. There was a significant regression between the activity rate and the number of termites present in a wood trap block, with a minimum detectable number of ≈50 workers per liter of wood. The success of these field tests suggests that currently available acoustic systems have considerable potential to detect and monitor hidden infestations of termites in urban trees and around building perimeters in addition to their present uses to detect and monitor termite infestations in buildings.