Under the increase in atmospheric CO2 during the last century, variable increases in the intrinsic water-use efficiency (Wi), i.e., the ratio between carbon assimilation rate (A) and stomatal conductance (gs), of C3 vegetation have been observed. Here, we ask if long-term nutrient status and especially nitrogen supply have an effect on the CO2 response of Wi in a temperate seminatural C3 grassland. This analysis draws on the long-term trends (1915–2009) in Wi, derived from carbon isotope analysis, of archived hay and herbage from the Park Grass Experiment at Rothamsted (South-East England). Plant samples came from five fertilizer treatments, each with different annual nitrogen (N; 0, 48 or 96 kg ha−1), phosphorus (P; 0 or 35 kg ha−1) and potassium (K; 0 or 225 kg ha−1) applications, with lime as required to maintain soil pH near 7. Carbon isotope discrimination (13Δ) increased significantly (P < 0.001) on the Control (0.9‰ per 100 ppm CO2 increase). This trend differed significantly (P < 0.01) from those observed on the fertilized treatments (PK only: 0.4‰ per 100 ppm CO2 increase, P < 0.001; Low N only, Low N+PK, High N+PK: no significant increase). The 13Δ trends on fertilized treatments did not differ significantly from each other. However, N status, assessed as N fertilizer supply plus an estimate of biologically fixed N, was negatively related (r2 = 0.88; P < 0.02) to the trend for 13Δ against CO2. Other indices of N status exhibited similar relationships. Accordingly, the increase in Wi at High N+PK was twice that of the Control (+28% resp. +13% relative to 1915). In addition, the CO2 responsiveness of 13Δ was related to the grass content of the plant community. This may have been due to the greater CO2 responsiveness of gs in grasses relative to forbs. Thus, the greater CO2 response of grass-rich fertilized swards may be related to effects of nutrient supply on botanical composition.