We developed and tested a new method to collect CO2 from the surface to deep layers of a peatland for radiocarbon analysis. The method comprises two components: i) a probe equipped with a hydrophobic filter that allows entry of peat gases by diffusion, whilst simultaneously excluding water, and, ii) a cartridge containing zeolite molecular sieve that traps CO2 passively. We field tested the method by sampling at depths of between 0.25 and 4 m at duplicate sites within a temperate raised peat bog. CO2 was trapped at a depth-dependent rate of between ∼0.2 and 0.8 ml d−1, enabling sufficient CO2 for routine 14C analysis to be collected when left in place for several weeks. The age of peatland CO2 increased with depth from modern to ∼170 BP for samples collected from 0.25 m, to ∼4000 BP at 4 m. The CO2 was younger, but followed a similar trend to the age profile of bulk peat previously reported for the site (Langdon and Barber, 2005). δ13C values of recovered CO2 increased with depth. CO2 collected from the deepest sampling probes was considerably 13C-enriched (up to ∼+9‰) and agreed well with results reported for other peatlands where this phenomenon has been attributed to fermentation processes. CO2 collected from plant-free static chambers at the surface of the mire was slightly 14C-enriched compared to the contemporary atmosphere, suggesting that surface CO2 emissions were predominantly derived from carbon fixed during the post-bomb era. However, consistent trends of enriched 13C and depleted 14C in chamber CO2 between autumn and winter samples were most likely explained by an increased contribution of deep peat CO2 to the surface efflux in winter. The passive sampling technique is readily portable, easy to install and operate, causes minimal site disturbance, and can be reliably used to collect peatland CO2 from a wide range of depths.