The size and number of rosettes of Cirsium vulgare were censused in a 4 ha sheep grazing trial on lowland calcareous grassland in paddocks receiving controlled winter, spring and summer grazing treatments. Spring grazing significantly increased thistle rosette numbers, and there were always fewer rosettes in size classes 250mm-300 mm under the heavier grazing treatments. The emergence of seeds of Cirsium vulgare sown by hand into each of the grazing treatments was monitored and showed a positive effect of spring grazing. When the percent emergence of sown seeds was used as a covariate in the analysis of rosette numbers in the experimental paddocks, it was found to account for 77% of the variance in thistle numbers. It was concluded that the main effect of grazing on thistle rosette numbers is an indirect one exercised through the effects of grazing on germination conditions. Seeds of Cirsium vulgare were sown into artificial gaps at another grassland site, to determine the effect of gap-size upon seedling emergence. Germination was poor, but significantly more seedlings emerged in gaps 10-20 cm diameter, than in 5 cm gaps or in control plots without a gap. A computer simulation model was used to explore the relationship between gap density and thistle population dynamics. A threshold density of gaps was found to exist, below which thistle populations went extinct, and above which the thistle population grew geometrically. The degree of aggregation of dispersed seeds did not alter the threshold gap-density for plant extinction, but did affect the rate of increase of the thistle population when the threshold gap-density was exceeded. It is concluded that strategies for controlling Cirsium vulgare populations by grazing manangement will be most effective if aimed at reducing suitable sites for establishment in spring. The success of such attempts will depend upon the soil fertility of the site, and sudden outbreaks of Cirsium vulgare infestation can be expected if a sward gradually deteriorates through over-grazing.