We deployed an automated multiplexed soil-respiration (SR) system to monitor partitioned soil CO2 component fluxes (from roots, mycorrhizal hyphae and heterotrophs) in a UK grassland using a combination of shallow surface (total SR flux), deep (excluding roots and mycorrhizal fungi) and 20-µm pore mesh window soil collars (excluding roots only). Soil CO2 efflux was monitored during a 3-month period during summer. Repeated cutting of mycorrhizal connections in some of the mycorrhizal treatments enabled assessment of subsequent recovery of mycorrhizal fluxes and a comparison with deep collar fluxes. After soil collar insertion, fluxes in the deep collars were significantly reduced, by approximately 40%. Whereas fluxes in the uncut, mycorrhizal collar treatments remained close to those from the surface collar, cut mycorrhizal treatments showed an immediate reduction after cutting to values close to those from the deep collar with a subsequent recovery of around 4 weeks. Overall, the autotrophic root and mycorrhizal flux was relatively stable throughout. Whereas root fluxes contributed about 10–30% of the total flux during the initial larger flux period, this declined and there was an increased mycorrhizal contribution during the latter part of the measurement period. Moreover, SR flux components differed in their response to key climatic factors, with root fluxes responding equally to temperature and light. Importantly, whereas the heterotrophic flux component responded strongly to temperature and soil moisture, the mycorrhizal component responded much less to those factors, but more to light. We also investigated treatment impacts over time on soil biochemical variables such as microbial biomass C, extractable C, microbial quotient and metabolic quotient, and bacterial community structure, and discussed these in relation to measured SR fluxes and the partitioning technique.