Three-year field assessment of gene flow from the genetically modified herbicide-resistant spring wheat variety 'Andros' to its non-transgenic counterpart has been conducted. A circular field trial design where the wheat plants containing the bar and gfp genes were planted in a central plot, while the recipient non-transgenic plants were grown in eight compass sectors at a distance of 1-5 m from the pollen source, has been developed. Gene flow was analyzed by testing the glufosinate-based herbicide resistance in seedling progeny. The phenotypic and molecular examination of more than 712,000 germinated seeds allowed us to draw a conclusion that the pollenmediated transgene flow might occur at a low frequency (\0.8 %) even if the nontransgenic wheat was located in the proximity to its transgenic counterpart. The strong asymmetric distribution of gene flow and the maximum outcrossing rate were detected in compass sectors following the direction of the dominant wind. The gene flow rate averaged over all wind direction varied in different years from 0.134 to 0.416 %. The dramatic reduction in transgene flow frequencies was observed while increasing the distance between transgenic and non-transgenic plots. The rise of average gene flow due to significant increase of the transgene pollen area was not confirmed statistically. These results predict that the pollen-mediated transgene flow in spring wheat can be maintained at negligible levels while the existence of a short isolation distance in order to prevent contamination of adjacent non-GM wheat.