Knowledge of the effects of agricultural practices on weed seedbank dynamics is essential for predicting future problems in weed management. This article reports data relative to weed seedbank structure after 18 years of continuous application of conventional tillage (CT, based on mouldboard ploughing) or no tillage (NT) within three crop sequences (continuous wheat, WW; wheat-faba bean, WF; and wheat-berseem clover, WB). Tillage system did not affect the size of the total weed seedbank, but altered both its composition and the distribution of seeds within the soil profile. In particular, the adoption of CT favoured some species (mainly Polygonum aviculare), whereas the continuous use of NT favoured other species (Papaver rhoeas, Phalaris spp. and Lactuca serriola). The effects of tillage system on weed seedbank size and composition were less pronounced in the WB cropping system than in either the WW or WF. Compared with WF and WB, WW resulted in an increase in total weed seedbank density (about 16 000 seedlings m-2 in WW, compared with 10 000 and 6000 seedlings m-2 in WF and WB, respectively) and a reduction in weed diversity, with a strong increase in some species (e.g. Polygonum aviculare). Our results for the effect of NT application on weed seedbank size and composition suggest that farmers should only apply such a conservative technique within an appropriate crop sequence