Raphanus raphanistrum (wild radish) is a highly competitive weed in winter crops. Integrated weed management practices and decision-making tools benefit from the ability to predict seed longevity and dormancy status in the soil seed bank, as well as time and density of emergence in the field. We wondered if unique values taken from databases could serve for modelling purposes, whatever the origins of the weed populations. We investigated dormancy and longevity of fruits buried in the soil over a four-year seed burial experiment of two highly contrasted populations that differ by their biogeographical origin (oceanic versus continental climate), their habitat (arable field versus undisturbed) and their fruit size (thick versus thin fruit wall). High viability persisted over two years, followed by a rapid reduction, especially for the ‘continental-undisturbed-thin’ population. Dormancy cycling, with dormancy at its lowest in October, was observed for the ‘oceanic-arable-thick’ population, whilst a slow decrease without clear seasonal pattern was found for the other population. These results indicate different ways of regulating seed persistence in the soil, which might be taken into account when building demographic models. These differences could be due to interactions between fruit wall thickness and the other factors; it is possible that a thicker fruit wall increases longevity by promoting dormancy by physical restriction, but depending on temperature. Thinner fruit walls would make plants display other adaptive strategies to maximize survival time in the soil seed bank. Studies involving more populations and isogenic material are needed to address this specific question.