Grazers may bring about vegetation change in pastures through effects on the creation and colonization of gaps. The natural colonization of three sizes of artificial gap in a species-poor, fertile pasture was monitored in an experiment in which two seasons of sheep grazing were applied, each at two levels: spring (grazed or ungrazed) and summer (hard or light grazed). The experiment had a 2 × 2 factorial design with two blocks in eight paddocks. Gap filling was slow; after 50 weeks the ramet density of the smallest gaps was 71 % of that of the surrounding sward. Seedling establishment was the dominant mode of colonization, accounting for 59 % of colonizing ramets. The remaining colonists were from clonal ingrowth. Smaller gaps filled fastest, having higher clonal ramet densities than the larger gaps. Gap size had complex effects on seedling colonization. Over the whole gap area there was a non-significant trend for increased densities of seed-derived plants in smaller gaps, but in just the central area of the gap, increased gap size increased the density and size of seed-derived plants. There were no spring grazing or grazing treatment × gap size interactions. Harder summer grazing did not affect gap filling rates but it decreased seedling densities and increased clonal ramet densities, possibly by reducing flowering. Species differed in gap colonization ability, measured by the change in a species' frequency between the surrounding sward and the colonized gap. Species also differed significantly in the ratio of seed-derived to clonal colonizing ramets, ranging from almost complete clonal colonizers to almost complete seedling colonizers. Species with a higher proportion of colonizing ramets derived from seed had higher colonization abilities than more clonal species. Increased gap size increased the frequencies of some species and decreased those of others. Both spring and summer grazing treatments affected the frequencies in the gaps of some species. There were no interaction effects of gap size and grazing treatment on gap species composition. Treatment effects on species with high proportions of seed-derived ramets were due only to effect.r on the frequencies of seed derived ramets and treatment effects on more clonal species were due to effects on the frequencies of clonal ramets. These results show that grazing may bring about vegetation change by its effects on the rate of gap creation because colonized gaps have different species frequencies to the closed sward. However, these gap effects are complex; the frequencies of some common species were decreased in gaps but those of other common species were increased, and most rarer species were decreased in gaps. Gap size and grazing treatment effects show that the effects of gap creation on vegetation change will be dependent on the sizes of gaps created