Soil is the largest terrestrial carbon (C) store, containing an estimated ~1500 Gt C in the upper 1 m of soil. The long term storage of soil organic C (SOC) requires that it is somehow protected from microbial decomposition – or ‘stabilised’ – in the soil matrix. Three mechanisms are commonly identified as factors controlling the stability of SOM: chemical recalcitrance, physical protection in aggregates and adsorption to soil mineral surfaces. The stability of SOC in the soil matrix can be influenced by management practices and changes in soil structure can lead to loss of SOC and increases in greenhouse gas (GHG) emissions. It is, therefore, important to understand the impact that management practices have on SOC stability and to manage soils in such a way as to optimise the volume of SOC which is locked away for climatically significant periods of time. Two methods are generally used to estimate SOC stability: indirectly by measuring CO2 fluxes as a proxy for SOC microbial decomposition, or directly through physical fractionation of soil in to pools with different levels of physical and chemical protection. Both methods were employed in this thesis. Arable and grassland soils which represent the range of soil textures and climatic conditions of the main agricultural areas in the UK were incubated at two different moisture contents and with or without inorganic fertiliser application and GHG fluxes from them were monitored. Soil texture, mineral N concentration and soil C concentration were found to be the most important measured variables controlling GHG fluxes of the UK agricultural soils in this study. The results were generally in support of those found in the literature for a wide range of soils, conditions and locations; however, N2O emissions from the two Scottish soils appeared to be more sensitive to inorganic N fertilisation at the higher moisture content than the other soils, with the N2O emissions being exceptionally high in comparison. Although incubations of whole soils are useful in measuring the impacts of soil management practices on GHG emissions under controlled conditions they do not identify the mechanisms controlling the stability of SOC. Dividing SOM into functional pools may identify different C stabilising mechanisms and improves soil C models. A large number of operationally defined separation methods have been used to fractionate SOM into biologically meaningful pools of different stability. Direct comparisons of different fractionation methods using radiocarbon (14C) dating and spectroscopic analyses has not previously been undertaken. Average 14C ages and chemical composition of SOM fractions isolated from a grassland soil using three published and frequently applied fractionation methods were compared. (1) a density separation technique isolating three fractions (2) a combined physical and chemical separation isolating five fractions (3) a hot-water extraction method isolating two fractions. The fractions from Method 1 had the most distinct average 14C ages, the fractions from Method 2 fell into two age groups, and both Method 3 fractions were dominated by modern C. The average 14C ages of the labile fractions from Method 1 and 2 were higher than the mineral bound fractions, although they made up a relatively small proportion of the total SOC. This was a surprising result, and spectroscopic analysis confirmed that these fractions had greater relative contents of aliphatic and aromatic characteristics than the mineral bound fractions. The presence of black C in a whole soil sample and one of the labile fractions from Method 2 was confirmed by hydrogen pyrolysis. The availability of archived soils from an abandoned long term tillage treatment experiment and the ability to relocate the plots provided a unique opportunity to assess the resilience of SOC stocks to land management practices several years after the conversion from arable to grassland. SOC stability was assessed by soil fractionation of archived (1975) and freshly collected (2014) soil samples. The mass corrected SOC stocks from the four different treatments (deep plough, shallow plough, chisel plough and direct drill) were higher in 2014 than 1975 across the whole profile (0 – 36 cm). Reductions were observed at some depths for some treatments but the overall effect was an evening out of SOC stocks across all plots. The fractionations (using Method 2), revealed that there was a relative increase in the mass of the sand and aggregate fraction but a decrease in the relative proportion of SOC stored in this fraction (physically protected). There was also a significant increase in the C:N ratio of the silt and clay fraction (chemical adsorption). This suggests that reduced disturbance of agricultural soils leads to preferential physical stabilisation of fresh SOM but also increased adsorption of older material to mineral surfaces. The labile fractions were sensitive to land-use change in all tillage treatment plots, but were more sensitive in the low impact tillage plots (chisel plough and direct drill) than the inversion tillage plots (deep plough and shallow plough). It is well established that tillage disrupts aggregation. However, a direct measurement of the level of SOM physical protection in the soil matrix due to aggregation has not previously been undertaken. The soil was fractionated using Method 1 (fractions with distinctly different 14C ages) and isolated soil fractions were incubated separately, recombined and mixed in to whole soil at three different temperatures. The C respiration rate of the isolated intra-aggregate fraction was generally consistently as high as the whole soil. This supports the theory that there is a labile component of soil which is protected from decomposition by physical protection within aggregates. Therefore, the lack of any priming effect with the addition of labile fractions to the whole soil, and indeed the suppression of emissions relative to the whole soil, was unusual. Fractions and whole soils incubated at 25 and 35 °C had a wider range of Q10 (temperature sensitivity) values than those incubated at 15 and 25 °C, however, median values were surprisingly similar (range from 0.7 to 1.9). Overall, the results from this thesis highlight the importance of the soil structure in stabilising C. Disrupting aggregates leaves a proportion of otherwise stable C susceptible to loss through microbial decomposition, particularly when the entire soil matrix is disrupted. It also provided some unexpected results which warrant future investigation; in particular, further direct measurement of physical stabilisation of SOM in soils of different type, from different climates and different land uses would be useful.