In a chalk grassland in north-western France, vegetation and seed rain were studied along two transects. We compared the efficiency of three types of seed traps (funnel traps, sticky traps, pots exposed in the field) to estimate the annual seed rain and to reveal various processes involved in plant community regeneration. Data were analysed to compare seed density, species richness and composition across seed traps and vegetation. Geostatistical analyses (semivariograms. were used to detect possible autocorrelation and to examine patterns of spatial variation in seed rain and vegetation. The results show that (1) funnel traps are most efficient (56 species, 8079 seeds.m–2). (2) Each type of seed traps brings different information about the processes involved in seed dispersal and seedling establishment. Sticky traps help to understand import and export of seeds in a community through anemochory. Funnel traps may provide a measure of the auto-regeneration capacity of the vegetation through its local seed production. Pot traps integrate various processes involved between seed dispersal and seedling emergence. (3.) The exposition to environmental conditions generates marked spatial patterns in species richness measured in pots and in vegetation, selecting species able to survive. Pot traps indicate that numerous species producing seeds in standing vegetation are unable to establish in the field through seedling recruitment. Our study emphasizes the need of using different types of seed traps and spatial sampling configurations to assess the regenerative potential of standing vegetation and the various mechanisms involved in seed dispersal. A standardised protocol for seed rain sampling was finally developed< Réduire