The risk assessment of genetically-modified plants pursuant to Annex II B of EU Directive 94/15/EC assumes that it is possible to infer the environmental impacts of a crop plant from its characteristics, so most of Annex II should also be applicable to conventional plants. To test this, we surveyed reports on the ecological impacts of the cultivation of non-transgenic crop plants with novel or improved traits and, in three cases, investigated whether Annex II B would have been adequate to indicate the effects. Such an assessment appears to be feasible only if the time frame on which it is based is short, so that long-term effects cannot be assessed. Secondly, the plant must be genetically homogenous which is not always granted, e.g. with forest-trees. Thirdly, the cultivation area must be defined. Differences in the behaviour of foreign plants between their original and cultivation habitats may be ecologically relevant and should be assessed. In the (few) cases where direct inference of the observed effects was possible from inherent traits, these effects often correlated with poor adaptation to local environmental conditions. The ecological impacts of traits that had been introduced in order to overcome poor adaptation may differ widely according to the way in which the traits are exploited. In practice, the effects of agricultural measures are more important than the effects of gene transfer and invasiveness, although the latter currently play a major role in risk assessment. In the light of these deliberations, a modification of Annex II B of EU Directive 94/15/EC is suggested.