A B S T R A C TIn order to label the CO 2 injected underground for geological storage and allow it to be differentiated from natural sources, a panoply of additive chemical tracers have been proposed. Yet, the transport of these tracers relative to CO 2 in pore space is currently poorly constrained. This leads to uncertainty as to whether tracers will act as an early warning of CO 2 arrival, or be preferentially retained in the pore space making them ineffective. Here, we present the factors affecting transport of noble gases and SF 6 relative to CO 2 in a porous rock. Using a porous sandstone core, each of the tracers were loaded into a sample loop and injected as discrete gas pulses into a CO 2 carrier stream at five different experiment pressures (10-50 kPag upstream to ambient pressure downstream). Tracer arrival profiles were measured using a quadrupole mass spectrometer. Significantly, our results show that peak arrival times of helium were slower than the other noble gases at each pressure gradient. The differences in peak arrival times between helium and other noble gases increased as the pressure gradient along the system decreased and the curve profiles for each noble gas differ significantly. The heavier noble gases (Kr and Xe) along with SF 6 show an earlier arrival time and a wider curve profile compared to He and Ne curves through the CO 2 carrier gas stream. This shows that Kr and Xe could be substituted for SF 6 , a potent greenhouse gas, in tracer applications. For comparison, CO 2 pulses were passed through a N 2 carrier gas resulting in significantly slower peak arrival times compared to those of noble gases and SF 6 . Hence, all investigated tracers when co-injected with CO 2 could potentially act as early warning tracers of CO 2 arrival, though we find that Kr, Xe and SF 6 will provide the most robust advance warning. Analysis of our experimental results shows that they cannot be explained by a simple one dimensional flow model through a porous medium. We outline a conceptual model that incorporates different preferential flow paths depending on flow velocities of individual gas streams. This model can explain the observed dataset and shows that the flow of noble gases and SF 6 tracers is influenced by pore-scale heterogeneity.