A series of 6 daylight observations was made each summer and again each winter over 2 years to map cattle distribution on a California foothill pasture. Sixty animals were used in the study with no animals appearing in > 1 observation series. During daylight hours, small herds of cows containing between 14 and 16 animals were scan-sampled and videotaped every 15 minutes. A global positioning system was used to record the position of the camera to aid in accurately locating individual animals. Animal locations and individual identifications were then entered into a geographic information system (GIS) by on-screen digitizing using color orthophotographs. Animal positions were determined to be within 5 m of their true location. Association software, ASSOC1, was used to analyze animal positions to determine cattle subgroups and herd units. This position-based grouping was compared with observation-based grouping by researchers. Direct observation also identified dominant herd members. Older animals, up to 16 years of age, were generally dominant over younger animals, and subgroups tended to be composed of animals of similar age. The size of naturally occurring subgroups was between 3 and 6 animals. Some animals exhibited independence in their actions and behaviors compared with subgroup members. ASSOC1 produced grouping results consistent with direct observations. However, accurate interpretation of the ASSOC1 results depended on direct observational data. ASSOC1 identified close association patterns in 3 of the observations that defined the dominant animals in the herd. Forage availability and thermoregulatory needs influenced the distance between associated subgroup members. Distance between animals decreased when animals sought shade in summer or shelter in winter. Computer analysis of spatial data from GPS collars may be able to determine the social structure and identify dominant animals in herd situations. Incorporating knowledge of cattle social behavior should improve management of cattle on the range.