Drawing on field studies of pollen dispersal, we identify features of the hybridization process that need quantification. Our emphasis is on standardized measures, as opposed to the idiosyncratic and often anecdotal methods with which gene flow or out-crossing data are currently reported. In addition to proposing specific maximum likelihood approaches, we summarize some results to date from small-scale field trials that bear on the risks anticipated for large-scale commercialization. We conclude that absolute containment of recombinant pollen or genes is unlikely if physical isolation is the only containment strategy. Because we conclude that the escape of transgenic pollen is inevitable, we argue that the focus of risk analysis should be shifted towards the ‘invasiveness’ of transgenic plants and ‘mitigation’ of their impact on natural, as well as agricultural systems.