Communal roosting has been proposed as a mechanism for facilitating thermoregulation in bats. We examined the effects of cluster size on body temperature (Tb) and metabolism (measured as 02 consumption, Voa) in Noctilio albiventris, a colonial insectivorous Neotropical bat with an adult body mass of 33-45 g.Measurements were made for group sizes of one, two, three, four, eight, and sixteen individuals at ambient temperatures (Ta) of 100, 200, and 300 C. At 300 C, Vo, is minimal (0.016 mL/[g . min]) and is not significantly affected by group size. At 100 and 200 C, Vo, is considerably higher than at 300 C, but it declines rapidly as group size increases from one to four. Increasing the group size beyond four individuals has little additional effect on 1Vo2. By switching from solitary roosting to a cluster offour or more, individual bats can reduce their Vo2 from 0.082 to 0 051I mL/(g min) at a Ta of IO0C (a 38% savings) and from 0.048 to 0.025 mL/(g min) at a T, of 200 C (a 47% savings). Mean body temperature was lowest for groups offour at all Ta 's. At Ta 's of 100 and 200 C there was more variance in Tb and a greater tendency for low Tb in groups offour or more than in smaller groups. The minimum Tb in groups of one to three was 32.30 C, but we measured Tb's as low as 27. 40 C in larger groups. We conclude that clustering lowers the energy cost of roosting in N. albiventris because (1) heat loss rates are reduced in groups and (2) individuals in groups are more likely than solitary bats to further decrease metabolism by becoming hypothermic. Noctilio albiventris may be selected to avoid hypothermia when solitary because low Tb probably increases their vulnerability to predation (the minimum Tb for flight is 350 C). Presumably, risks ofpredation are lower for bats roosting in groups.