The ten-ten rule can be used to predict the community consequences of releasing GMOs. This, combined with data on the feral state of British crop plants, predicts that almost all GMOs will become at least casual, that more than 10% will establish, and about 10% of those will become pests. Which constructs will fall into which category cannot be predicted from their characters. In particular, Baker characters have no predictive value for weediness. The fourteen most troublesome invasive plant pests show a remarkably diverse set of characters. GMOs, being mainly derived from crop plants, and in some cases with genes that are likely to enhance survival, can be expected to have an appreciable effect on nonagricultural ecosystems, once a range of different constructs have been released. Familiarity is unlikely to be an effective defence against new ecological effects.