Transgenic oilseed rape (Brassica napus L.; OSR) is estimated to be environmentally and economically problematic because volunteers and ferals occur frequently and because of its hybridisation potential with several wild and weedy species. A proposed mitigation strategy aims to reduce survival, in particular in conventional OSR crops, by coupling the transgenic target modification with a dwarfing gene to reduce competitive fitness. Our study allowed us to access potential ecological implications of this strategy.On a large scale (>500 km(2)), we recorded phenological and population parameters of oilseed rape plants for several years in rural and urban areas of Northern Germany (Bremen and surroundings). The characterising parameter were analysed for differences between wild and cultivated plants.In rural areas, occurrences of feral and volunteer OSR together had an average density of 1.19 populations per square kilometre, in contrast to urban areas where we found 1.68 feral populations per square kilometre on average. Throughout the survey, the vegetation cover at the locations with feral OSR ranged from less than 10% to 100%. Our investigations gave clear empirical evidence that feral OSR was, on average, at least 41% smaller than cultivated OSR, independent of phenological state after onset of flowering.The findings can be interpreted as phenotypic adaptation of feral OSR plants. Therefore, it must be asked whether dwarfing could be interpreted as an improvement of pre-adaptation to feral environments. In most of the sites where feral plants occurred, germination and establishment were in locations with disturbed vegetation cover, allowing initial growth without competition. Unless feral establishment of genetically modified dwarfed traits are specifically studied, it would not be safe to assume that the mitigation strategy of dwarfing also reduces dispersal in feral environments.With respect to OSR, we argue that the proposed mitigation approach could increase escape and persistence of transgene varieties rather than reducing them. We conclude that the development of effective hazard mitigation measures in the risk evaluation of genetically modified organisms requires thorough theoretical and empirical ecological analyses rather than assumptions about abstract fitness categories that apply only in parts of the environment where the plant can occur.