In New Zealand, nitrogen (N) inputs and losses have increased dramatically following establishment (and subsequent intensification) of European style pastoral agriculture from the 1800s. An indicator that could identify soils which are most vulnerable to N loss would be useful to help target management practices aimed at reducing unwanted N losses. The natural abundance of ¹⁵N relative to ¹⁴N (δ¹⁵N) in soils is one potential indicator of N loss, because during most N transformations in soils, ¹⁴N is preferentially processed and lost (e.g. via ammonia volatilisation, denitrification or nitrate leaching). Therefore the overarching hypothesis for this thesis was that pastoral soils under intensive management regimes (with high N inputs, cycling and losses) would become progressively enriched with ¹⁵N relative to soils under less intensive management. This hypothesis was tested by measur ing δ¹⁵N in soils from four forest-to-pasture chronosequences, and in archived soils from six long-term (4–57 year) grazed field trials with different fertiliser or irrigation regimes. Three of the forest-to-pasture chronosequences were on pumice soils where pine forests had been converted to dairy pastures. The fourth chronosequence was on a podzol soil in Northland, where native scrub had been converted to sheep grazed pastures. Surface soil δ¹⁵N on the pumice soil increased significantly from pine forests (2 ‰) to long-term pastures (4.1 ‰). In contrast, there was no clear relationship between pasture age and soil δ¹⁵N for the chronosequence on the podzol soil. The Northland soil displayed extreme podzolization and weathering, and had been previously disturbed by gum diggers, implying results could be relatively unique to this soil. The two longest field trials were at Winchmore on the Canterbury plains, where different rates of superphosphate and irrigation had been applied for ~50 years. Soil δ¹⁵N increased more in treatments receiving higher rates of superphosphate or more frequent irrigation, and there were significant positive correlations between the average rate of change in soil δ¹⁵N, and total pasture production, clover production (thus N fixation) and calculated N losses. Soil δ¹⁵N was also measured in archived soils from three long-term (¹⁵–25 year) superphosphate trials on North Island hill country. In these trials there were no consistent differences in soil δ¹⁵N between treatments. The shorter duration of the hill country trials (compared to the Winchmore trials), combined with smaller differences in pasture production between treatments and higher variability due to complex topography, may have contributed to the lack of observed differences between treatments. Indeed, slope and aspect did have a significant influence on soil δ¹⁵N with higher values on sheltered east facing slopes, and on easy slopes than steep slopes. The final trial studied was a N fertiliser trial, in which N rates ranging from 0 to 750 kg ha–¹ y–¹ were applied to hill country pastures over a 4 year period. Soil δ¹⁵N increased significantly with time in treatments receiving >100 kg N ha–¹ y–¹, and the increase was more rapid as N rate increased. There was also a positive correlation between the rate of change in soil δ¹⁵N and nitrate leaching (p<0.001). In general, results from this thesis showed that soil δ¹⁵N under intensively managed pastures (i.e. those receiving higher rates of fertiliser or irrigation) was higher than under less intensively managed pastures. It was concluded that higher soil δ¹⁵N in the more intensively managed pastures was most likely due to the influence that fertiliser or irrigation had on pasture production, N fixation by clover, and the flow on effects this had on animal stocking rates and N cycling and isotope fractionating loss processes. However, results from the hill country superphosphate trials and the chronosequence on the podzol soil, demonstrated that other factors (such as slope and aspect) can overwhelm or suppress the expected increases in δ¹⁵N in some situations. Therefore soil δ¹⁵N will probably be most accurate as an indicator of long-term management intensity, and management induced N losses, at sites with the same (or similar) topography, soils and climate. Average rates of change in soil δ¹⁵N over the duration of the trials investigated in this thesis, ranged from –0.007 ‰ y–¹ to 0.35 ‰ y–¹, with rates of change being <0.1 ‰ y–¹ in all trials except the N rate trial (where N inputs were very high). This suggests that under ‘typical’ pastoral management regimes, any changes in bulk soil δ¹⁵N will probably only be detectable at decadal time scales. A preliminary indicator of past management intensity and N inputs and losses was proposed for New Zealand soils, based on surface soil δ¹⁵N values. (1) Soils with δ¹⁵N values <3 ‰ will be from extensively managed pastures, forests, or other natural ecosystems with low N inputs and losses. (2) Soils with intermediate δ¹⁵N values (3–5 ‰) will be from sites which have been under moderate management intensity, with moderate N inputs and losses. (3) Soils with δ¹⁵N values >5 ‰ will be from sites which have been intensively managed for a number of decades and subject to high N inputs and losses (e.g. N inputs >100 kg ha–¹ y–¹). More research will be required to further test the usefulness of this indicator. Key areas for future research include: gaining a better understanding of the effect of topography and soil type on soil δ¹⁵N, and making more direct measurements of the isotopic composition of different N inputs and outputs to more clearly identify mechanisms driving changes in soil δ¹⁵N in pastoral systems.