While herbicide use in agriculture is expected to have many effects on surrounding weed communities, it is largely unknown how plant exposure to sublethal doses of herbicide may subsequently impact plant-pollinator interactions. We tested the hypothesis that sublethal herbicide exposure indirectly alters plant-pollinator interactions through changes in plant traits, and specifically through alterations in floral display. Using a common garden experiment, we exposed 11 weed species to the herbicide dicamba and examined the potential for changes in pollinator abundance and patterns of pollinator visitation as well as alterations to plant traits. We found variation among plant species in the extent of damage from dicamba drift, and variation in size, flowering time, and flower displays, with some plant species showing negative impacts and others showing little effect. Pollinator frequencies were reduced in dicamba-exposed plots, and pollinator visits were reduced for some weed species yet not for others. Structural equation modeling revealed that the relationship between flower display and pollinator visits was disrupted in the presence of dicamba. Our study provides the most comprehensive picture to date of the impacts of herbicide drift on plant-pollinator interactions, with findings that highlight an underappreciated role of services supplied by weedy communities.