Home range behavior and spatial relationships of chases were examined for differences that might indicate which sex-age classes of Columbian ground squirrels (Spermophilus columbianus) would likely be infanticidal. Locations of individually marked squirrels were recorded at regular intervals and at the start of chases on 2-3 days per week during the 1971 active season. Location maps were used to measure home range size, home range overlap, distances between individuals, and distances between intruders and adult female home burrows. The probability of an adult female chasing another adult female decreased as the distance of the intruder from the owner's burrow increased, but was not related to the distance between the two adult females. Conversely, the probability of an adult female chasing members of other sex-age classes decreased as the distance between the squirrels increased, but was not related to distance of the intruder to the female's burrow. The latter results were also obtained for chases by adult males of members of all sex-age classes. Overlap between home ranges of adult females was significantly less when infants were in their burrows compared to gestation and post-lactation periods. However, overlap of adult female home ranges by those of other sex-age classes did not change seasonally. These results suggest that only adult females are likely to be infanticidal, which agrees with the two observations so far reported. They do not support Sherman's (1981) prediction that yearling males will also be infanticidal. Among the Marmotini, infanticide has been observed in Marmota flaviventris (Steiner 1972; Armitage et al. 1979; Brody and Melcher 1985), Cynomys ludovicianus (Hoogland 1985), Spermophilus beldingi (Sherman 1981), S. parryii (McLean 1983), S. columbianus (Balfour 1983; Waterman 1984), and S. beecheyi (Trulio et al. 1986). Circumstantial evidence (e.g. cannibalism, aggression) suggests it probably occurs in other members of the ground squirrel genus Spermophilus (Michener 1982; Sherman 1982). It may be a widespread but low frequency behavior among the Marmotini in general. Field documentation is difficult because the actual killing may occur below ground. Infanticide in the Marmotini conforms to two of the five hypotheses for the adaptiveness of infanticide reviewed by Hausfater and Hrdy (1984). Some individuals, most commonly adult females, rarely eat their victims. They may be increasing future reproductive success by reducing competition for burrows or other resources that they, or their own offspring, will need. Other infanticidal individuals, most often males, eat at least part of the dead juveniles. They may be getting protein important for rapid growth, overwintering survival, and future breeding success. Both hypotheses may apply for some individuals. Both of the reported observations of infanticide in Columbian ground squirrels (S. columbianus) involved resident adult females that had litters of their own (Balfour 1983; Waterman 1984). Sherman (1981), however, predicted that males of this species might be infanticidal, as are those of arctic ground squirrels (S. parryii), because the social systems of the two species are similar. Although direct observations of infanticide are mandatory to clearly understand this behavior in ground squirrels, differences in the ways parents interact with other individuals can suggest which individuals in a species pose a threat of infanticide. The