Soil-borne nitrous oxide (N2O) emissions have a high spatial and temporal variability which is commonly attributed to the occurrence of hotspots and hot moments for microbial activity in aggregated soil. Yet there is only limited information about the biophysical processes that regulate the production and consumption of N2O on microscopic scales in undisturbed soil. In this study, we introduce an experimental framework relying on simplified porous media that circumvents some of the complexities occuring in natural soils while fully accounting for physical constraints believed to control microbial activity in general and denitrification in particular. We used this framework to explore the impact of aggregate size and external oxygen concentration on the kinetics of O2 consumption, as well as CO2 and N2O production. Model aggregates of different sizes (3.5 vs. 7 mm diameter) composed of porous, sintered glass were saturated with a defined growth medium containing roughly 109 cells ml-1 of the facultative anaerobic, nosZ-deficient denitrifier Agrobacterium tumefaciens with N2O as final denitrification product and incubated at five different oxygen levels (0-13 vol-%). We demonstrate that the onset of denitrification depends on the amount of external oxygen and the size of aggregates. Smaller aggregates were better supplied with oxygen due to a larger surface-to-volume ratio, which resulted in faster growth and an earlier onset of denitrification. In larger aggregates, the onset of denitrification was more gradual, but with comparably higher N2O production rates once the anoxic aggregate centers were fully developed. The normalized electron flow from the reduced carbon substrate to N-oxyanions (edenit-/etotal- ratio) could be solely described as a function of initial oxygen concentration in the headspace with a simple, hyperbolic model, for which the two empirical parameters changed with aggregate size in a consistent way. These findings confirm the important role of soil structure on N2O emissions from denitrification by shaping the spatial patterns of microbial activity and anoxia in aggregated soil. Our dataset may serve as a benchmark for constraining or validating spatially explicit, biophysical models of denitrification in aggregated soil. © 2018 Schlüter, Henjes, Zawallich, Bergaust, Horn, Ippisch, Vogel and Dörsch.