With the commercial development of the global positioning system (GPS), it is now possible to monitor the distribution of free ranging cattle and derive measures to describe landscape use. Animal GPS data can be integrated with a geographic information system (GIS) detailing topography, vegetation, soil type and other landscape features. Combining GPS and GIS information is useful for understanding how animals respond to spatial variability. This study quantified land-type preferences for Brahman cross steers over three time periods, from October 2004 to March 2006 in a replicated trial, under heavy (4 ha/AE; animal equivalent of ~450 kg steer) and light (8 ha/AE) stocking in four, ~105 ha paddocks of subtropical semi-arid savanna near Charters Towers, Queensland, Australia. The grazing trail was conducted at a scale much less than would be found in commercial situations. Consequently, the spatial pattern of cattle reported here may not represent what occurs at a commercial scale and implications are discussed. Results were analysed in terms of the spatial distribution of steers fitted with GPS devices in each of the four paddocks and for each stocking rate to provide insight into cattle distribution and land-type preferences. Steers walked in excess of 6 km per day, regardless of stocking rate, and exhibited diurnal patterns of movement, with peak activity around dawn (0500–0700 hours) and dusk (1800–2000 hours). The spatial distribution of the collared steers was not uniform and appeared to be strongly influenced by the prevailing drought conditions and location of water points within each paddock. A hierarchy of drivers for distribution was identified. With the exception of drinking water location, land subtype based on soil-vegetation associations influenced animal distribution. Preference indices (ŵi) indicated that steers selected sites associated with heavy clay and texture contrast soils dominated by Eucalyptus coolabah Blakely & Jacobs (ŵi = 5.33) and Eucalyptus brownii Maiden & Cambage (ŵi = 3.27), respectively, and avoiding Eucalyptus melanophloia F.Muell. ridges (ŵi = 0.26) and Eucalyptus cambageana Maiden (ŵi = 0.12) on sodosols. The results suggest that spatial variation in cattle distribution within a paddock may be more critical than overall stocking rate in influencing the pattern of biomass utilisation. However, to quantifying the effects of different grazing land management practices on animal distribution on a commercial scale, additional studies in extensive paddocks are required.