Technical Abstract: Animal and land costs usually restrict the number of treatments, replicates, and animals per paddock or pasture in grazing trials. Because of limited resources, grazing trials on research stations often are conducted using small paddocks with a few animals per paddock. Land and animal restrictions can be reduced by conducting trials on farms using animals provided by cooperating farmers. Farmers typically want only a single replicate on their farms. Farms can be used as replicates but concerns about statistical precision and the ability to detect treatment differences has restricted the use of this design. In the Central Great Plains, USA, both types of grazing trials have been conducted. The statistical precision of an on-farm grazing trial in which farms were used as replicates with replicated paddock trials on research stations was compared. Standard error of the mean for both average daily gain and gain per hectare were lower in the on-farm trial than in any of the replicated small paddock trials. Because of their larger size, paddocks in on-farm trials could be stocked by more animals which reduced the experimental variation for animal gain and gain per hectare. Well managed on-farm trials in which farms are used as replicates can provide greater statistical precision than small paddock trials on research stations because larger paddocks and greater numbers of animals can be used.