Undesirable grazing distribution results in land degradation and inefficient forage utilization. Rewards like food supplements have been commonly provided at predictable times and locations in the less preferred areas in order to improve grazing distribution. The problem with this approach is the generation of search patterns that are concentrated at certain times and locations, causing either overgrazing of new areas or the rapid return of the animals to the most preferred areas. Our model of spatial–temporal schedule of reinforcement proposes that rewards that are unpredictable in space and time should promote search patterns that are extended in space and time. In order to test predictions from the model, we studied how the spatial–temporal arrangement of supplementation influenced selection of feeding sites by sheep on a grass-legume pasture. Each experimental plot had a fertilized and an unfertilized side (8 m × 16 m each) either adjacent (Trial 1) or separated by an alley (1 m × 32 m) (Trial 2). In both trials groups of three ewes were randomly assigned to each of four treatments resulting from the combination of two spatial and two temporal arrangements of supplementation in the unfertilized side of the pasture. All treatments received the same amount and type of supplement (500 g alfalfa/corn pellets per ewe per day). There were two repetitions per treatment. Animal positions were recorded every 2 min during 2 h daily grazing sessions for 12 (Trial 1) or eight (Trial 2) consecutive days. The response variable was the proportion of time spent on the unfertilized side of the pasture. In Trial 1 neither treatment effects nor the interaction treatment by day was significant (P > 0.05). In Trial 2, the random spatial and/or temporal arrangements of supplementation tended to increase the occupation of the unfertilized side of the pasture by sheep. The differences among treatments were larger at the beginning of the trial, when the availability of forage was relatively higher than towards the end of the trial, when the availability of forage was relatively lower. In both trials the proportion of time spent on the unfertilized side of the pasture decreased (P < 0.01) as the daily grazing session progressed from the first half hour to the second half hour to the last hour. Our results raise some interesting possibilities that warrant future tests of the model of spatial–temporal schedule of reinforcement to modify selection of feeding sites by free grazing animals. Fil: Distel, Roberto Alejandro. Universidad Nacional del Sur. Departamento de Agronomía; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - Bahía Blanca. Centro de Recursos Naturales Renovables de la Zona Semiárida. Universidad Nacional del Sur. Centro de Recursos Naturales Renovables de la Zona Semiárida; Argentina Fil: Soca, P. M.. Facultad de Agronomía ; Uruguay Fil: Demment, M. W.. University of California at Davis; Estados Unidos Fil: Laca, E. A.. University of California at Davis; Estados Unidos