Off-target contamination (or spray drift) during agricultural chemical application, arising from removal of small (diameter < 100 m) droplets from sprays by atmospheric or vehicle-generated cross flows, is investigated experimentally. The primary requirement for realistic wind tunnel studies is identified from background review as matching the full-scale logarithmic man velocity profile with suitable surface roughness parameters. A general calculation scheme is presented for spacing horizontal flat plates to simulate weakly-sheared mean velocity profiles. Adequate full-scale matching of logarithmic mean velocity profiles is achieved after systematic equipment modification. Comparative field and wind tunnel experiments using single nozzles show adequate agreement following the above approach, indicating that air entrained into the liquid spray stabilizes the spray to the cross flow action. Measurements within an agricultural spray in still air show that small droplets are passively transported within the entrained air field, whose characteristic turbulence length scale is too small to contribute to droplet dispersion. Wind tunnel studies employing conventional sprays show small droplet removal associated with regions where the entrained air velocity is less than the cross flow velocity, with essentially passive downwind transport. Numerical simulations of spray drift must clearly incorporate characteristics of the entrained air velocity field.