Different types of mixing cell models are analysed in detail. One type is found to be second-order accurate when reactions are not considered but reduces to first-order accuracy when a sorption reaction is taken into account. The second type is only first-order accurate even when a reaction is not considered. The most commonly used mixing cell model is extended to account for different surface boundary conditions, including that for a landfill. The exit conditions for a finite spatial domain were taken into consideration by making use of boundary layer corrections. It is found that the results obtained using the mixing cell model coupled with a non-linear isotherm are very accurate when compared with results obtained from a Crank-Nicolson scheme. It was observed that the accuracy of the results is dependent on the spatial step size. Best results are obtained when the spatial step size is close to twice the dispersivity. Application of the mixing cell model with a boundary layer correction was demonstrated by comparison with experimental data from a laboratory Na-Ca exchange experiment in which a solution containing Na was passed through a column filled with Ca-saturated loamy sand. The mixing cell model was also applied to breakthrough curves of two organic compounds (carbontetrachloride and tetrachloroethylene) observed in a field experiment conducted at Borden, Ontario, Canada. The mixing cell model described satisfactorily the Na transport in loamy sand and also the transport of organics in a Borden aquifer.