To assess gene dispersal from oilseed rape volunteers, the whole life cycle of deliberately broadcast seeds and seed losses during harvest was observed in four different tillage operations. Treatments I and 3 were immediate stubble tillage, later followed by primary tillage with a plough or cultivator. In treatment 2, the stubble tillage was delayed for four weeks and later followed by primary tillage ploughing. Treatment 4 was zero tillage. The following crop in all treatments was winter wheat. In autumn, between about 7 and mostly less than 50% of the initial rape seeds emerged, whereas volunteer emergence was less than 0.01% in spring. Depending on the treatment, a seed bank was built up reaching up to 30% of the initial number of seeds. The seed bank was largest when seeds were immediately incorporated into the soil by stubble tillage. Up to 85% of seeds could not be found; the heaviest losses occurring when the seeds remained on the soil surface for a while in treatments 2 and 4. When volunteers flowered at the same time as an oilseed rape crop, gene dispersal in space by pollen transfer was possible. Flowering volunteer plants could be observed mainly in treatments 3 and 4, at a population density of up to 0.78 plants m(-2) in treatment 3. A new generation of seeds could be produced by the volunteers that may found a further cohort of volunteers and enable gene dispersal in time or, in case of volunteers emerging in a rape crop, be harvested with the crop.