Regression analysis was used to examine the importance of organic nitrogen (%N), soil water content (θv), soil pH, and C: N ratio for predicting N mineralisation in a small field plot. Undisturbed soil cubes (c. 1·7 cm3) were collected from the soil surface and received treatments of drying and rewetting, urea, substrate derived from clover leachate, or no amendment, and were incubated at either –10 or –30 kPa for 20 days. The data confirm the hypothesis that within a small field plot, θv and %N explain most of the variation in net N mineralisation and nitrification. The pore size classes of 0·6–10 and 10–30 µm made disproportionately small and large contributions to N mineralisation, respectively, apparently due to non-uniform distribution of organic N through the pore system. When soluble N substrate was added to the soils, both these pore classes appeared to support mineralisation. We concluded that prior to sampling, the microbial biomass had been more active in the pores 0·6–10 µm, and had nearly exhausted the organic substrates in this pore class, whereas this was not so for the 10–30 µm pore class. Drying and rewetting increased the importance of %N as a predictor of N mineralisation, probably because this treatment disrupted physical protective mechanisms of organic N. Soil pH was generally not a useful predictor of N mineralisation and often seemed to be a dependent rather than an independent variable in relation to nitrification. Neither was C: N ratio a useful predictor of N transformation processes, and this was probably related to physical regulatory mechanisms in the soil.