The Mt. Stafford region, central Australia, provides well-exposed metasedimentary rocks whose protoliths were deposited at ca. 1.84–1.81 Ga, and that were metamorphosed at greenschist to granulite facies pressure–temperature conditions at 1.8 Ga. The higher-grade rocks show evidence of low-pressure partial melting (2–4 kbars) over a temperature range of 650° to 850 °C. These exposures afford evaluation of the extent to which N is retained through devolatilization and partial melting reactions experienced in the shallow- to mid-continental crust. In the Mt. Stafford metapelitic/metapsammitic suite, whole-rock δ 15N air values of the lowest-grade, greenschist facies rocks (Zone 1) range from + 2.3 to + 6.2‰ (mean = + 3.6‰; 1σ = 1.5‰; n = 7) and values for the highest-grade granulite facies rocks (Zone 4) range from + 2.1 to + 8.5‰ (mean = + 5.2‰; 1σ = 2.1‰; n = 20). Six of the twenty samples at the highest grade have values higher than those of the greenschist rocks regarded as likely protoliths. All but a few of the Zone 4 rocks have N concentrations lower than the mean N concentration for the lowest grade rocks (ranges of 107–346 ppm, mean 254 ppm, for Zone 1 rocks, and 43–361 ppm, mean 185 ppm, for Zone 4 rocks). For the higher grade rocks, which experienced multiple partial melting reactions, varying retention of N, as NH 4 + in K-feldspar or as N 2 in cordierite, may indicate that N is held in peritectic products of partial melting rather than in the melt. Although assessments of element loss in the Mt. Stafford suite are greatly complicated by the large degree of protolith-related heterogeneity, some elements do show modest decline in concentration, with increasing grade, consistent with their loss during devolatilization and partial melting. Nitrogen appears most similar in behavior to Rb (the LILE most similar in ionic radius to NH 4 +) but, of these two elements, N shows somewhat greater evidence for whole-rock loss. K 2O shows only very subtle hints of whole-rock loss, in some samples, and Cs shows considerably greater loss relative to N and Rb (resulting, for example, in increases in Rb/Cs at the higher grades). Increased whole-rock Ba/Rb for some higher-grade samples, relative to that for lower-grade equivalents, likely reflects the formation of a major Ba reservoir in K-feldspar at the expense of biotite (resulting in no obvious whole-rock Ba loss), with Rb liberated from biotite breakdown partitioning into the melts that at least partly left the system. Reduced whole-rock B concentrations at the higher grades resulted from the breakdown of tourmaline during the melting reactions. These varying degrees of reduction in whole-rock N, Rb, Cs, and B (and possibly K 2O) concentration are accompanied by decreased concentrations in H 2O (as indicated by LOI), Li, and U, the loss of U resulting in increased Th/U in the highest-grade rocks. Apparent differential loss of these elements, based on the whole-rock data, could reflect the amounts of melt produced and the extents to which the melts were removed at the scales sampled by the hand-specimens that were analyzed. Six of the twenty Zone 4 samples showing greater shift in δ 15N, to values of + 7.4 to + 8.4‰, tend to be among the samples showing the largest changes in other major and trace element concentrations and ratios believed to reflect melt removal (e.g., K 2O, Cs/TiO 2, B/Al 2O 3, Th/U, and ratios among Rb, Sr, Ba, and Cs). The metasedimentary rocks at Mount Stafford also contain 217 to 1186 ppm C, likely as reduced C (graphite) but, at the higher grades, probably also as CO 2 in the channels of cordierite. At greenschist grade, whole-rock δ 13C V-PDB values are relatively uniform, with organic compositions (mean = − 25.1‰; 1σ = 1.6‰; n = 7), but at the highest grade (Zone 4) values are as high as − 17.7‰ (mean = − 22.2‰; 1σ = 2.3‰; n = 8). Overall, C and N, apparently largely of organic origin, show surprising retention in rocks having experienced peak temperatures of > 800 °C, demonstrating their compatibility in mineral phases in fluid and melt residues (for N, as NH 4 + in K-feldspar and as N 2 in cordierite; for C, as graphite and as CO 2 in cordierite). These data have implications for deep-Earth N cycling, bearing on the size of the N reservoir in the mid-continental crust (and thus long-term storage of N previously residing in the atmosphere and biosphere), and for the use of the δ 15N of upper-crustal intrusive rocks as a tracer of their deep crustal sources.