For most insects, size determines the call frequency. This paper describes the acoustics of a small brachypterous cricket (Rufocephalus sp.; body length 9.6 mm) producing a call with a carrier frequency of approximately 3.2 kHz from a subterranean burrow. Crickets such as Gryllus campestris are approximately twice this length and produce a call frequency close to 5 kHz. The burrow of Rufocephalus opens via a small hole with a diameter of 3.2 mm. The neck of the hole at approximately 1.4 mm depth opens to a vertical two-part burrow with an upper vase-shaped chamber 16.1 mm in height with a diameter of 9.4 mm. This top chamber connects via a 6.4mm high (diameter 5.2 mm) neck to a more irregular chamber approximately 18 mm high with a width of approximately 11 mm. The walls of the top chamber neck and of the upper part of the lower chamber are smooth and appear to be sealed with saliva. The song has a mean centre frequency of 3.2 kHz and is made up of variable-length trills of pulses of mean duration 15.8 ms. Many song pulses had smooth envelopes and their frequency did not vary by more than +/-40Hz from the centre frequency, with a relative bandwidth Q-3dB of over 50. Other pulses showed considerable amplitude and frequency modulation within the pulse. When driven by external sound, burrows resonated at a mean frequency of 3.5 kHz with a mean quality factor Q of 7.4. Natural-size model burrows resonated at similar frequencies with similar Q values. One cricket, which had previously called from its own burrow at 2.95 kHz, sang at 3.27 kHz from a burrow that resonated at the same frequency. Life-size model burrows driven by external sound resonated at similar frequencies to the actual burrows; models three times life size resonated at one-third of this frequency. In all models, the sound pressure was more-or-less constant throughout the top chamber but fell rapidly in the neck of the burrow; the phase of the sound was effectively constant in the top chamber and neck and fell through approximately 180 degrees in passing from the neck into the lower chamber. A numerical model of the sound flow from region to region gave essentially similar results. A resonant electrical model fed from a high-impedance source with discrete tone bursts at different frequencies showed similar amplitude and frequency modulation to the various types of song pulses that were observed. It is suggested that the high purity of the songs results from close entrainment of the sound-producing mechanism of the insect's wings to the sharply resonant burrow.