Global positioning system (GPS) data collected over a 4-yr period on 52 crossbred young cows grazing a 146-ha pasture were used to determine whether cattle establish patch-scale rotational patterns within pastures. Cow positions at 5-min intervals were recorded during 20 d in late winter/early spring. Estimated per capita forage allowance (PCFA) was 347 kg herbage · cow⁻¹, 438 kg herbage · cow⁻¹, 1 104 kg herbage · cow⁻¹, and 1 884 kg herbage · cow⁻¹ in 2004, 2005, 2006, and 2007, respectively. Cumulative winter/early spring precipitation (CPPT) was low in 2004 and 2006 (35 mm and 30 mm, respectively) and high in 2005 and 2007 (119 mm and 112 mm, respectively). Structured query language codes developed for this study were used to 1) select grazing GPS points with movement velocities between 1 m · min⁻¹ and 20 m · min⁻¹, 2) overlay location data on a pasture map subdivided into 30 × 30 m pixels, and 3) calculate percentage of grazed pixels (% GP), pixel residence time (RT), revisit rate (RR), and return interval (RI) for each animal. Cows grazed 31% ± 5.9 SEM of all pixels for 21 min ± 3.7 SEM, visited grazed pixels 1.6 times ± 0.18 SEM, and returned to grazed pixels after 5 D ± 2  SEM. As PCFA increased, % GP decreased (r = −0.42) and RI increased (r = 0.73) significantly (P < 0.01); however, RT decreased (r = −0.46) and RR increased (r = 0.6) significantly (P < 0.01) with increasing CPPT. Pixel attributes (elevation, aspect, slope, percentage of tree cover, and distance from water, roads, and fences) failed to explain variation in pixel RT (R² = 0.28) regardless of PCFA. The same predictors explained most of the variation in pixel RR and RI when PCFA was high (R² = 0.86 and R² = 00.76, respectively). Cows appear to establish their own patch-scale rotational patterns within pastures. Nonforage pixel attributes appear to have a strong influence on such patterns.