Many researchers have proposed that the stimulus of plant growth under elevated [CO2] observed in short-term experiments will be moderated in the longer term by a reduction in soil nitrogen (N) availability linked to decreased litter quality and/or increased litter production. However, these negative feedbacks may be offset to some extent by a stimulus in N fixation linked to increased root exudation. The aim of this modelling study is to examine how changes in litter quality/quantity and root exudation - if they occur - will affect the CO2 responses of net primary productivity and ecosystem carbon (C) storage on different timescales. We apply a model of C and N cycling in forest ecosystems (G'DAY) to stands of Norway spruce (Picea abies, L. Cast) growing at a N-limited experimental site at Flakaliden, Sweden, and draw the following conclusions: (1) in the absence of changes in litter quality and root exudation, the short-term CO2 stimulus of litter quantity leads to only a minimal CO2 stimulus of productivity or C storage in the medium term (approximate to 20 years) and long term (approximate to 200 years), because of constraints on soil N availability; (2) increasing plant nitrogen use efficiency (via a decrease in the N:C ratio of new litter) makes little impact on these results; (3) a significant CO2 response in the medium term requires a substantial decrease in the N:C ratio of older litter, when it is approaching stabilisation as soil organic matter, although the long-term CO2 response remains small; and (4) an increase in N fixation leads to a small effect on productivity in the short term, but a very large effect on both productivity and C storage in the long term. These results suggest that soil N constraints on the long-term CO2-fertilisation effect can be overcome to a significant extent only by increases in N acquisition, although only modest increases may be required.