δ15N signals in plant and soil material integrate over a number of biogeochemical processes related to nitrogen (N) and therefore provide information on net effects of multiple processes on N dynamics. In general little is known in many grassland restoration projects on soil–plant N dynamics in relation to the restoration treatments. In particular, δ15N signals may be a useful tool to assess whether abiotic restoration treatments have produced the desired result. In this study we used the range of abiotic and biotic conditions provided by a restoration experiment to assess to whether the restoration treatments and/or plant functional identity and legume neighborhood affected plant δ15N signals. The restoration treatments consisted of hay transfer and topsoil removal, thus representing increasing restoration effort, from no restoration measures, through biotic manipulation to major abiotic manipulation. We measured δ15N and %N in six different plant species (two non-legumes and four legumes) across the restoration treatments. We found that restoration treatments were clearly reflected in δ15N of the non-legume species, with very depleted δ15N associated with low soil N, and our results suggest this may be linked to uptake of ammonium (rather than nitrate). The two non-legume species differed considerably in their δ15N signals, which may be related to the two species forming different kinds of mycorrhizal symbioses. Plant δ15N signals could clearly separate legumes from non-legumes, but our results did not allow for an assessment of legume neighborhood effects on non-legume δ15N signals. We discuss our results in the light of what the δ15N signals may be telling us about plant–soil N dynamics and their potential value as an indicator for N dynamics in restoration.