The open pollination of maize raises the possibility of outcross of conventional and organic varieties with adventitious pollen from a genetically-modified source field. Analysis of potential outcross requires knowledge of atmospheric flow and subsequent pollen dispersal. The flow fields are often complex. When small maize plots are surrounded by a shorter crop such as soybean, the step function in surface characteristic substantially alters the flow within and near the maize field. Tall field-edge borders also affect the flow, with the potential of reducing pollen dispersal away from the source field. This three paper study explores the dispersal of pollen from maize fields, as well as the effect of a tall field-edge border on flow and dispersal.;For this purpose, a three dimensional Lagrangian particle dispersion (LPD) model has been adapted for maize pollen dispersal and validated with field studies from Iowa in August 2003 and 2005. Millions of particle trajectories were simulated following a random flight approach; each particle is tracked and the resulting pollen cloud is proportional to the ensemble of particles. The flow fields required to drive the LPD were provided by scaling laws valid within the atmospheric surface layer (paper one) or by three dimensional implementation of the Wang and Takle (WT) shelterbelt model (papers two and three). The shelterbelt model provides physically consistent flow fields and allows better analysis of the sensitivity of pollen dispersal to flow characteristics.;The dispersion model generates a realistic pollen deposition plume for the 2003 cases, similar in shape to the observations (paper one). Using flow fields from scaling relationships to drive the LPD, for the case of maize surrounded by soybean, the model explains most of the observed location to location pollen deposition variability. The model doesn't produce enough downwind deposition far from the field, due in part to the lack of realistic vertical velocity. Using WT flow fields and releasing much more pollen also produces realistic deposition patterns, but there is an over-prediction of pollen far from the field (paper two). Since the model produces realistic deposition patterns, it is a useful assessment tool to explore the sensitivity of dispersal to flow properties. Sensitivity tests revealed that the maize field reduces pollen transport downwind of the field due to a reduction in wind speed, while the altered turbulence and vertical velocity tend to slightly increase dispersal away from the field.;The observed effect of a field-edge border was explored in paper three. With the addition of a sorghum border at field edge, there is modestly less observed long-range deposition. The simulations produced realistic spatial patterns of particle deposition that included the sharp near-source deposition gradient and longer-range dispersive tail. However, the model produced too little pollen at mid distances from the field (∼30 to 50 m) and over-predicted pollen at larger distances (>50 m), with a dispersive plume that appears too wide. Some of this far-field error may be due to differences in sampling, since the model employs a larger count region than the pollen traps. Comparing the simulated border and control deposition, the chief differences are a modest increase in within-field deposition and a slightly faster downwind decrease in deposition with distance for the border field. This results in increased downwind pollen deposition for the control field. The simulated effect of the border on long-range dispersal is small in absolute terms, but large in relative terms (with up to a 50% reduction at some downwind locations). This suggests that field-edge borders may reduce pollen dispersal enough to reduce the likelihood of outcross with neighboring fields. The model sensitivity to border characteristics is modest, even when the border height is increased by 1 m. On the other hand, a simulation with upwind flow imposed over the domain results in decreased within-field deposition and substantially increased pollen within ∼150 m of the field. This indicates that the maize field itself has a substantial sheltering impact.