Elevated nitrogen (N) deposition can affect litter decomposition directly, by raising soil N availability and the quantity and quality of litter inputs, and indirectly by altering plant community composition. We investigated the importance of these controls on litter decomposition using litter bags placed in annual herb based microcosm ecosystems that had been subject to two rates of N deposition (which raised soil inorganic N availability and stimulated litter inputs) and two planting regimes, namely the plant species compositions of low and high N deposition environments. In each microcosm, we harvested litter bags of 10 annual plant species, over an 8-week period, to determine mass loss from decomposition. Our data showed that species differed greatly in their decomposability, but that these differences were unlikely to affect decomposition at the ecosystem level because there was no correlation between a species' decomposability and its response to N deposition (measured as population seed production under high N, relative to low N, deposition). Litter mass loss was ∼2% greater in high N deposition microcosms. Using a comprehensive set of measurements of the microcosm soil environments, we found that the most statistically likely explanation for this effect was increased soil enzyme activity (cellobiosidase, β-glucosidase and β-xylosidase), which appears to have occurred in response to a combination of raised soil inorganic N availability and stimulated litter inputs. Our data indicate that direct effects of N deposition on litter input and soil N availability significantly affected decomposition but indirect effects did not. We argue that indirect effects of changes to plant species composition could be stronger in natural ecosystems, which often contain a greater diversity of plant functional types than those considered here. © 2007 Elsevier Ltd. All rights reserved.