We present a model of solute transport that explains the large-scale behavior of the solute tracer-test plumes at the Macrodispersion Experiment (MADE) site as the result of advection and rate-limited mass transfer between mobile and small-scale immobile domains. This model does not consider the process of dispersion and yet provides an alternative explanation of the evolution of the observed concentration profiles. Compared to the macrodispersion model, the mass transfer model better represents the change in mobile dissolved mass with time, the peak of the concentration profile, and the profile asymmetry. Specifically, unlike the macrodispersion model, the mass transfer model explains the facts that the observed mass of the plume was greater than the injected mass in early snap shots of the plume and less than the injected mass at late times. We suggest that the injected mass advects through the mobile domain and diffuses into and out of the immobile domain. The immobile domain consists of a combination of low-permeability zones on the scale of centimeters to decimeters (the Darcy-scale immobile domain), and intragranular porosity, dead-end pores, and surface sorption (the pore-scale immobile domain). We suggest that the mobile domain was sampled preferentially when water was extracted. Therefore, at early times, relatively clean water in the immobile domain was not sampled and incorrectly assumed to contain high solute concentrations. Similarly, the mass at late times was underestimated because solute trapped in the pore-scale immobile domain was not extracted during sampling and therefore ignored. The combination of advection and slow mass transfer is consistent with the fact that the peak of the plume migrated only ~5 m by the termination of the experiment, as well as the different behavior of bromide and tritium tracers. ¿ 2000 American Geophysical Union