Interactions between plants and herbivores often vary on a geographic scale. Although theory about plant defenses and tolerance is predicated on temporal or spatial variation in herbivore damage, no single study has compared the pattern of herbivory, plant defenses and tolerance to herbivory of a single species across a latitudinal gradient. In 2002–2005 we surveyed replicate salt marshes along the Atlantic coast of the United States from Florida to Maine. At each field site we scored leaves of Iva frutescens for herbivore damage. In laboratory experiments we measured constitutive resistance and induced resistance in I. frutescens from high and low latitude sites along the Atlantic Coast. In another common garden experiment we studied tolerance to herbivory of I. frutescens from various sites. Theory predicts that constitutive resistance should matter more when damage is high, and induced resistance when herbivory is high but variable. In the field, average levels of herbivore damage, and spatial and temporal variation in herbivore damage were all greater at low versus high latitudes, indicating that constitutive as well as induced resistance should be stronger at low latitudes. Consistent with this prediction, constitutive resistance to herbivory was stronger at low latitudes. Induced resistance to herbivores was also stronger at low latitudes: it was deployed faster and lasted longer. Theory also predicts that tolerance to herbivory should be greater where average herbivory damage is greater; however, tolerance to herbivory in Iva did not depend on geographic origin. Our results emphasize the value of considering multiple ways in which plants respond to herbivores when examining geographic variation in plant–herbivore interactions.