AbstractRates of leaf extension and senescence were measured over three‐day intervals on neighbouring sets of caged and uncaged tillers on several occasions during summer 1983, both when grazing animals were present and also when they were absent. The weather was dry and drought conditions developed as the season advanced.Comparison of data from caged and uncaged tillers indicated that some consistent differences occurred. Loss of information through grazing (affecting 30% of uncaged, larger than average, tillers), contributed to the significantly reduced estimates of leaf extension of ungrazed, uncaged tillers compared with caged tillers. Senescence rates were higher in the open plot than under the cages; old leaves were grazed on only 8% of uncaged tillers so that data loss was less important as a cause of bias. In the drought conditions prevailing, it was thought that caging in the presence of grazing animals resulted in reduced moisture stress and contributed to the above results.Comparisons of caged and uncaged tillers when grazing animals were absent were made to examine the environmental effects of caging and the slight difference in tiller marking procedure. No differences were found in leaf extension rates, though a trend towards higher senescence rates for the uncaged populations was evident. It was concluded that the criteria limiting the use of cages (a period of protection less than the average interval between both leaf appearance and repeated defoliation) should be extended to include a time limit set by the need to avoid bias associated with the environmental effects of caging and cessation of grazing. To define this time limit better in particular circumstances, the extent and direction of bias in the estimation of leaf extension and senescence rates as affected by period of caging, tiller handling procedures, sward and climatic conditions requires further investigation.In the case of vegetative tillers, estimates of leaf extension rates can be made considerably more precise by using tiller size variables (pseudostem length, total green leaf length, leaf width) as covariates. A suggestion for the number of tillers to be measured per transect is given. However, no such increase in precision can be made for estimates of senescence rate