Questions Do short-term seed-addition experiments reliably tell us about seed limitation of perennial species? Does seed provenance affect recruitment success?Locations: Three dry grassland sites (Negrentino, Poma, San Giorgio), southern Switzerland.Methods: 18 000 seeds of Bromus erectus of three provenances were sown in a reciprocal design with temporal replication. Frequent checks were made of all emerged individuals of two seed cohorts over three years and continued annual checks of one cohort until year eight. Performance was determined by number and size of emerging and surviving individuals and a combined measure of population vigour.Results: From a three-year perspective, all sites appeared to be seed limited, and differences due to seed provenance were small. Over five years, two sites showed an increasingly superior performance of the local over the foreign populations. At one of these two sites, average individuals remained small and a decrease in number and individual phytomass since the fifth year suggested complete microsite limitation. Reproductive individuals only occurred at the other sites and confirmed seed limitation after eight years. The best performing site had the first reproductive individuals in the fourth year.Conclusions: B. erectus showed a long pre-reproductive phase during which the prediction of establishment success by individual counts can be misleading if plant size is not also measured. The effect of seed provenance was clearly indicated where populations established most successfully and local-over-foreign superiority increased with time.