[en] Highlights: • The aim was to investigate the effect of herbicides in an agro-ecosystem setting. • The levels of glyphosate in the field were estimated by measuring a dye tracer. • The deposition of glyphosate decreased rapidly with increasing distance. • Glyphosate had a significantly negative effect on the cumulative number of flowers. • Glyphosate did not affect the flowering time on any of the species significantly. Recent studies have shown that sub-lethal doses of herbicides may affect plant flowering, however, no study has established a direct relationship between the concentrations of deposited herbicide and plant flowering. Here the aim was to investigate the relationship between herbicide spray drift deposited on non-target plants and plant flowering in a realistic agro-ecosystem setting. The concentrations of the herbicide glyphosate deposited on plants were estimated by measuring the concentration of a dye tracer applied together with the herbicide. The estimated maximal and average deposition of glyphosate within the experimental area corresponded to 30 g glyphosate/ha (2.08% of the label rate of 1440 g a.i./ha) and 2.4 g glyphosate/ha (0.15% label rate), respectively, and the concentrations decreased rapidly with increasing distance from the spraying track. However, there were not a unique relation between distance and deposition, which indicate that heterogeneities of turbulence, wind speed and/or direction can strongly influence the deposition from 1 min to another during spraying. The effects of glyphosate on cumulative flower numbers and flowering time were modelled using Gompertz growth models on four non-target species. Glyphosate had a significantly negative effect on the cumulative number of flowers on Trifolium pratense and Lotus corniculatus, whereas there were no significant effects on Trifolium repens, and a positive, but non-significant, effect on number of flowers on Cichorium intybus. Glyphosate did not affect the flowering time of any of the four species significantly. Lack of floral resources is known to be of major importance for pollinator declines. The implications of the presented results for pesticide risk assessment are discussed.