ABSTRACTThe problems of designing, conducting and interpreting the results of grazing trials to evaluate tropical pasture mixtures are discussed in relation to an experiment wbich measured the effect of three fixed stocking rates upon two grass/legume pastures. A uniformity trial prior to grazing indicated considerable soil variability, which necessitated ground replication. Sequential grazing of three pasture replicates at a fixed stocking rate and a fixed interval of grazing resulted in sites of low fertility being heavily penalized. Such a form of replication was of value for measuring the effect of treatments upon sward composition, and, to a lesser extent, variation in animal production, within the experiment area. Differences in animal production between pasture species were small when compared with the effect of various stocking rates, emphasizing the need for an objective assessment of stocking rate when evaluating pasture species. Fixed‐stocking‐rate designs have more general application in the tropics because they are easier to conduct, and the results are repeatable and can be extrapolated to farm pract