A hypothesis is put forward that the long-lasting inducible responses of trees to herbivores, particularly lepidopteran defoliators, may not be active defensive responses, but a by-product of mechanisms which rearrange the plant carbon/nutrient balance in response to nutrient stress caused by defoliation. When defoliation removes the foliage nutrients of trees growing in nutrient-poor soils, it increases nutrient stress wich in turn results in a high production of carbon-based allelochemicals. The excess of carbon that cannot be diverted to growth due to nutrient stress is diverted to the production of plant secondary metabolites. The level of carbon-based secondary substances decays gradually depending on the rate at which nutrient stress is relaxed after defoliation. In nutrient-poor soils and in plant species with slow compensatory nutrient uptake rates the responses induced by defoliation can have relaxation times of several years. The changes in leaf nitrogen and phenolic content of mountain birch support this nutrient stress hypothesis. Defoliation reduces leaf nitrogen content while phenolic content increases. These responses of mountain birch to defoliation are relaxed within 3-4 years.