In Europe, regulatory thresholds restrict adventitious GM (genetically modified) presence in conventional crops. Minimum distances for the spatial separation of fields are often recommended to reduce field-to-field cross-pollination to an acceptable level. Field trials are typically the basis for setting separation distances. However, using records of wind direction and speed from weather stations across Europe, we predict theoretically that field-to-field windborne cross-pollination in maize, oilseed rape, sugar beet, and rice varies greatly according to the relative orientation of the GM and non-GM fields. Furthermore, at a given site and orientation from a GM field, we predict that the cross-pollination rate varies substantially from year to year. Consequently, even replicated field trials may inaccurately estimate typical levels of cross-pollination and therefore distort our perception of the separation distances required to achieve sub-threshold adventitious GM presence. We propose methods to predict the likely range in levels of cross-pollination based on the limited data typically available from field trials. Additionally, we suggest suitable time lags between peak flowering in adjacent fields that could be introduced to reduce cross-pollination to a specified level.