The transport and attenuation of cadmium (Cd) and rhodamine WT (RWT) in a pumice sand aquifer media was investigated using column experiments to study a scenario of point-source contamination. A pore-water velocity of 1.7-1.8 m/day, which is a typical field groundwater velocity in a pumice sand aquifer system, was applied to triplicate columns. A pulse of a solution containing Cd and RWT, together with the conservative tracer tritiated water ( 3 H 2 O) at pH = 7, was introduced into the columns. Experimental results showed that concentration breakthrough curves (BTCs) of 3 H 2 O were symmetrical and fitted well into an equilibrium model. In contrast, BTCs of Cd and RWT were asymmetrical with significant tailings and fitted well with a two-site adsorption/desorption model. The symmetric 3 H 2 O BTCs suggest that physical nonequilibrium was absent in the experimental system, therefore the asymmetrical BTCs of Cd and RWT were attributed to chemical non-equilibrium. Modelling results showed that, in comparison with 3 H 2 O, Cd was apparently retarded by 101-108 times in pumice sand aquifer media (apparent adsorption coefficient 7.33-9.24 ml/g) and underwent a mass loss of 20-30% that was probably because of precipitation of CdCO 3 . As CdCO 3 is extremely insoluble, Cd precipitation would be irreversible and therefore it would not contribute to the tailing of the Cd BTCs. The experimental results suggest that the adsorption and desorption of Cd in pumice sand aquifer media in hydrodynamic conditions was a kinetic process. Cd desorption rates were two orders-of-magnitude slower than its adsorption rates. This resulted in a prolonged mean residence time for Cd in pumice sand aquifer media, which was 10-12 days in the 18-cm-long columns under a flow velocity of 1.7-1.8 m/day. Since the mean residence time is only indicative for the arrival of the central of mass in a contaminant BTC, the time required for the total disappearance of Cd will be much longer than the mean residence time because of the significantly long tailing of the BTC. This implies that natural attenuation of Cd from a contaminated pumice sand aquifer would take a time period from decades to centuries. Batch isotherm experiments were also carried out to obtain Cd adsorption coefficients at equilibrium conditions. In comparison with the column results obtained from non-equilibrium conditions, adsorption coefficients of 20 ml/g obtained from the batch equilibrium experiments were 2-3 times higher for Cd. This finding suggests that care should be taken when using batch adsorption isotherm results to predict field problems because the delay in the appearance of contaminants in drinking water wells and springs could be overestimated. In comparison with 3 H 2 O, RWT was retarded 5-7 times and had BTCs that were significantly more spread out. Hence the use of RWT to indicate groundwater flow in pumice sand aquifers will underestimate groundwater velocity and overestimate aquifer dispersivity. About 4-14% of RWT mass was lost during its transport, prob...