Nycteris thebaica (Chiroptera: Nycteridae) is widely distributed in savannah habitat in Af 1983), and several published accounts deal with its roosting and feeding habits (Kingdon, LaVal, 1980) and echolocation calls (Fenton et al., 1983). Nycterids, including N. thebaica low aspect ratio and wingloading, and large wingspans, tail membranes, and ears (Norbe 1987), combinations of characters suggesting slow flight and high maneuverability. These s confirmed experimentally by Aldridge and Rautenbach (1987) who noted that the broadba calls of N. thebaica were another feature that would permit the bats to forage in cluttered Anecdotal field observations (Kindgon, 1974) and observations of animals in captivity (Fen suggest that members of this species sometimes hunt from perches, taking volant or nonv that these bats regularly use feeding perches (Felten, 1956;Fenton, 1975;LaVal and L purpose of this study was to examine certain aspects of the roosting and foraging behavior under natural conditions. We documented roost selection, vocalizations associated with roosting behavior in habitats of different structure.We conducted this study between 3 November and 5 December 1987 in Mana Pools N Zimbabwe (15*44'S; 29*21'E), and observations reported here were made within 2 km of pa and 800 m of the southern bank of the Zambezi River. Roost selection was studied by examinin of roosts selected by N. thebaica and the temperature regimes of these sites. The dimension roosts were measured to the nearest 0.1 m, and we used a Cole Parmer Digi-sense Model 852 to measure temperatures to the nearest 0. I*C. Temperatures from within roost cavities wer ambient temperatures in the shade outside roosts. Most roost temperatures were measured not occupied by bats but, for comparison, on three occasions we obtained readings from bats.Vocalizations of N. thebaica were recorded simultaneously on two tracks of a Racal Store 4D tape recorder operated at 76 cm/s, by use of two QMC S200 bat detectors (QMC Instruments Ltd., 229 Mile End Road, London El 4AA, United Kingdom). One microphone was located as close to the bats as possible, the other 2 m behind it. Vocalizations were recorded in a roost inside an unused military bunker, and inside a 3.6by 3.6-by 2-m screen tent with 1-mm-mesh-nylon walls. The frequency response of the recording system was calibrated before and after recordings and was flat to within 10 dB between 5 kHz and 100 kHz. During recording sessions, we monitored the amplitude of signals produced by the bats with a Tektronix 212 oscilloscope.