The effects of global N enrichment on soil processes in grassland ecosystems have received relatively little study. We assessed microbial community response to experimental increases in N availability by measuring extracellular enzyme activity (EEA) in soils from three grasslands with contrasting edaphic and climatic characteristics: a semiarid grassland at the Sevilleta National Wildlife Refuge, New Mexico, USA (SEV), and mesic grasslands at Konza Prairie, Kansas, USA (KNZ) and Ukulinga Research Farm, KwaZulu-Natal, South Africa (SAF). We hypothesized that, with N enrichment, soil microbial communities would increase C and P acquisition activity, decrease N acquisition activity, and reduce oxidative enzyme production (leading to recalcitrant soil organic matter [SOM] accumulation), and that the magnitude of response would decrease with soil age (due to higher stabilization of enzyme pools and P limitation of response). Cellulolytic activities followed the pattern predicted, increasing 35-52% in the youngest soil (SEV), 10-14% in the intermediate soil (KNZ) and remaining constant in the oldest soil (SAF). The magnitude of phosphatase response did not vary among sites. N acquisition activity response was driven by the enzyme closest to its pH optimum in each soil: i.e., leucine aminopeptidase in alkaline soil, beta-N-acetylglucosaminidase in acidic soil. Oxidative enzyme activity varied widely across ecosystems, but did not decrease with N amendment at any site. Likewise, SOM and %C pools did not respond to N enrichment. Between-site variation in both soil properties and EEA exceeded any treatment response, and a large portion of EEA variability (leucine aminopeptidase and oxidative enzymes), 68% as shown by principal components analysis, was strongly related to soil pH (r = 0.91, P < 0.001). In these grassland ecosystems, soil microbial responses appear constrained by a molecular-scale (pH) edaphic factor, making potential breakdown rates of SOM resistant to N enrichment.