A prototype field-deployable acoustic insect flight detector was constructed from a noise-canceling microphone coupled to an off-the-shelf digital sound recorder capable of 10 h recordings. The system was placed in an urban forest setting 25 times over the course of the summer of 2004, collecting 250 h of ambient sound recordings that were downloaded to a personal computer and used to develop detection routines. These detection routines operated on short segments of sound (0.093 s, corresponding to 4096 samples at 44100 Hz). A variety of approaches were implemented to detect insect flight tones. Simple approaches, involving sensing the fundamental frequency (1st harmonic) and 2nd harmonic, were capable of detecting insects, but generated large numbers of false positives because of other ambient sounds including human voices, birds, frogs, automobiles, aircraft, sirens, and trains. In contrast, combining information from the first four harmonics, from the interharmonic regions, and from the sound envelope, reduced false positives greatly. Specifically, in the 250 h of recordings, 726 clear insect buzzes were detected by the final algorithm, with only 52 false positives (6.5%). Running the final algorithm with all criteria liberalized by 20% increased the number of clear insect buzzes by 8%, to 784, but increased false positives to 471 (28% of total detections). The potential of using this approach for detecting mosquito activity using low-cost sensors is discussed.