There is increasing evidence that risk of predation is an important determinant of where animals choose to forage. Hoary marmots (Marmota caligata) concentrate their foraging in meadows close to sleeping burrows and that contain escape sites (Holmes, 1984). Yellow-bellied marmots (Marmota flaviventris) forage in areas that provide opportunity for predator detection (Carey, 1985). When given a choice, house sparrows (Passer domesticus) choose foraging locations closest to cover (Grubb and Greenwald, 1982). For many species, foraging decisions incorporate risk factors. Group size, a social variable, is known to affect risk of predation. Earlier detection of predators is one of the best documented ways in which protection by groups is afforded (Bertram, 1980; Hoogland, 1979). Also, because less vigilance is required for each individual, more time is available for foraging (Caraco, 1979; Holmes, 1984). Therefore, individuals that live or feed in groups not only have a smaller probability of being preyed upon, they may be able to obtain more or better quality food. Thus, social factors also can affect foraging indirectly. The present study was undertaken to determine the role of social factors in foraging behavior of a colonial mammal, the black-tailed prairie dog (Cynomys ludovicianus). Prairie dogs, like other foragers, may experience a conflict between foraging and predation risk (Covich, 1976). The quality and variety of food may vary with distance from a burrow, but greater distances pose an increasing risk, as the burrow is the prairie dog's main defense against predators (King, 1955). To minimize this conflict, prairie dogs, like other animals (Carey and Moore, 1986), could assess variables that relate to risk of predation, like group size, and adjust their behavior accordingly. For example, prairie dogs alter vigilance time as a function of colony size (Hoogland, 1979). Perhaps they might use group size to regulate how far away from a burrow to forage. During an 11-month period (September 1983-July 1984), I observed the distance animals foraged from burrows, and changes in group size (density). I expected that foraging distance from a burrow would increase as density increased. Observations were made at Martin Park Nature Center in Oklahoma City, Oklahoma. The prairie-dog town was established 4 years before the study and covered an area of about 0.4 ha. During the study period, the population reached a maximum of 37 animals. The vegetation in the town was about 75% mixed grasses (dominated by Cynodon dactylon and Schizachyrium scoparium) and 25% forbs (primarily Conoyza ramocissima). The area was mowed periodically; it was bordered by tall, unmowed Johnson grass (Sorghum halepense). Because the entire nature center was fenced, the main predators on this population were hawks (Buteo sp.). Alarm calling, which may indicate perceived risk, occurred during 6.5% of the observations. Observations were made with binoculars from a tower (3.6 m high) concealed in foliage 37 m from the edge of the town. Every 10 min, the distance of each animal from the nearest burrow entrance was estimated by counting the number of body lengths of an average adult prairie dog, about 33 cm (King, 1955). A value of 1 denoted animals at least one body length away and so on, to four or more body lengths. Four was the maximum distance that could be judged accurately, although animals often foraged farther away. A 0 designated that an animal was less than one body length from the burrow entrance. This value was assigned to animals sitting on a mound. The number of animals above ground also was recorded at each 10-min interval. Observations were discarded when an alarm call occurred just before or during the data-collection period. Data from animals engaged in chases, mound and burrow repair, or the gathering of nest materials were eliminated. Foraging is the most common behavior away from the burrow (King, 1955). With the exceptions noted all activities were assumed to be foraging. Weekly observations were made from 0840-1030 h. Median distances of all animals were calculated for each 10-min interval, resulting in 170 data points from 45 h of observations. Data were analyzed by linear regression and analysis of variance. The independence of the observations was uncertain because subjects were not marked; therefore, the degrees of freedom were reduced to the number of individuals in the population at that time. There was a significant positive correlation between animal density and foraging distance from burrows (Fig. 1). As the number of animals visible above ground increased, individuals were more likely to move farther away from a burrow entrance (r = 0.50, d.f. = 36, P < 0.01). This suggests that foraging distance was related to group size.