The importance of understanding the role of atmospheric conditions in pollen dispersal has grown in recent years with increased field-testing of genetically modified (GM) crop plants. An atmospheric model was used to characterize wind trajectories at 10 m and 100 m above GM pollen source fields located within a 4452-ha "control" area north of Madras, Oregon, USA, designated by the Oregon Department of Agriculture (ODA). The area was used in 2003 for the growth of GM creeping bentgrass (Agrostis stolonifera) engineered to be resistant to glyphosate herbicide. The presence of the GM gene (CP4 EPSPS) provided a distinct selectable marker for pollen-mediated gene flow to sentinel and resident Agrostis spp. plants. Linkage of GM gene presence with wind flow characteristics over the "control" area became essential to understand the timing and processes leading to long-distance transport of this pollen. Wind trajectories showed a general pattern of northwest to southeast air movement. Trajectory travel distances calculated hourly from 06:00 hours to 15:00 hours during the 2003 pollination period (15 June-15 July) showed movement up to 15 km from the "control" area's center by the first hour. Maximum travel distances increased to 40 and 55 km after two and three hours from release, respectively. Calculated wind trajectory positions corresponded with observed long-distance pollen-mediated gene flow in the seedlings of sentinel and resident plants. The highest correlations were found during the late morning hours. Back-calculated wind trajectories from sentinel and resident locations with GM-gene-positive progeny suggested that most successful fertilizations occurred in the direction of prevailing winds during late June 2003. The occurrence of positive progeny from sentinel plants, upwind of the "control" area during this period, indicated the additional influence of local topography on pollen dispersal.