The steering responses of three species of field crickets, Teleogryllus oceanicus, T. commodus, and Gryllus bimaculatus, were characterized during tethered flight using single tone-pulses (rather than model calling song) presented at carrier frequencies from 3-100 kHz. This range of frequencies encompasses the natural songs of crickets (4-20 kHz, Fig. 1) as well as the echolocation cries of insectivorous bats (12-100 kHz). The single-pulse stimulus paradigm was necessary to assess the aversive nature of high carrier frequencies without introducing complications due to the attractive properties of repeated pulse stimuli such as model calling songs. Unlike the natural calling song, single tone-pulses were not attractive and did not elicit positive phonotactic steering even when presented at the calling song carrier frequency (Figs. 2, 3, and 9). In addition to temporal pattern, phonotactic steering was sensitive to carrier frequency as well as sound intensity. Three discrete flight steering behaviors positive phonotaxis, negative phonotaxis and evasion, were elicited by appropriate combinations of frequency, temporal pattern and sound intensity (Fig. 12). Positive phonotactic steering required a model calling song temporal pattern, was tuned to 5 kHz and was restricted to frequencies below 9 kHz. Negative phonotactic steering, similar to the 'early warning' bat-avoidance behavior of moths, was produced by low intensity (55 dB SPL) tone-pulses at frequencies between 12 and 100 kHz (Figs. 2, 3, and 9). In contrast to model calling song, single tone-pulses of high intensity 5-10 kHz elicited negative phonotactic steering; low intensity ultrasound (20-100 kHz) produced only negative phonotactic steering, regardless of pulse repetition pattern. 'Evasive', side-to-side steering, similar to the 'last-chance' bat-evasion behavior of moths was produced in response to high intensity (greater than 90 dB) ultrasound (20-100 kHz). Since the demonstration of negative phonotactic steering did not require the use of a calling song temporal pattern, avoidance of ultrasound cannot be the result of systematic errors in localizing an inherently attractive stimulus when presented at high carrier frequencies. Unlike attraction to model calling song, the ultrasound-mediated steering responses were of short latency (25-35 ms) and were produced in an open loop manner (Fig. 4), both properties of escape behaviors.(ABSTRACT TRUNCATED AT 400 WORDS)