The maintenance of nitrogen limitation in terrestrial ecosystems remains a central paradox in biogeochemistry. Although plants that form a symbiotic association with nitrogen fixing bacteria should be at a competitive advantage over non-fixing plant species in N limited environments, N 2 fixing plants are uncommon in most mid-to high-latitude ecosystems. Theory and observation suggest that preferential grazing on N-rich tissues by herbivores, resource limitations to growth, reproduction and N 2 fixation, and temperature limitations to the activity of the N 2 fixing enzyme nitrogenase, explain the rarity of N 2 fixing plants. These ideas, however, have never been confronted by multifactor experiments in the field. In a 3 year field experiment, we found that the abundance, growth, reproductive output and fraction of plant-N derived from N 2 fixation in temperate, old-field ecosystems was constrained by the availability of phosphorus (P). Although the availability of light was crucial to the performance of old-field N 2 fixing plants, the largest gains in biomass and the rate of N 2 fixation were observed in the plots fertilized with P. By contrast, herbivory had no effect on the abundance, biomass and activity of N 2 fixing plants and inconsistent effects on foliar nitrogen concentrations (opposing directions, depending upon year), suggesting that herbivores do not affect the ecology of N 2 fixing plants in old field ecosystems, at least not over the course of 3 years. Together with a recent study demonstrating that C limitation explains the absence of N 2 fixing trees in temperate forests our analysis suggests that stand replacing disturbances shift the limitation on the abundance and activity of N 2 fixing plants from P early in secondary succession to light later in succession, as the forest canopy closes and incident light levels decline precipitously.