This study examines the dynamics of colloid mobilization and leaching from macroporous soil columns by means of laboratory experiments and numerical modeling. On the basis of a previous column study involving high and low water flow rates in structured soil, we designed a novel experiment emphasizing the time-dependence of the colloid release process. Intact macroporous soil columns were exposed to variable pauses in irrigation (flow interruption for 30 min, 1 d, or 7 d) followed by resumed infiltration. The experiments showed that (i) there was a seemingly unlimited source of in situ colloids even after prolonged leaching and (ii) the peak concentration of colloids in the effluent after the flow interruption increased with increasing length of the preceding pause. The results demonstrated that colloid mobilization is not controlled by hydrodynamic shear induced by the flowing water but is a time-dependent and possibly diffusion-limited process. We developed a simple, equivalent macropore model to investigate the hypothesis that colloid release to the flowing water is governed by two diffusion processes, one in a uniform water film lining the macropore and one in the crust of the macropore. The model was capable of reproducing and explaining the characteristic results of our soil column experiments and required no recalibration of exchange process parameters to simulate the particle mobilization after a flow interruption. However, model calibration yielded unexpected results with respect to the size of the diffusion coefficient of the crust and did not warrant accepting the dual diffusion model hypothesis. Using a simpler mass transfer concept to quantify the mobilization of colloids in 21 soil columns, we found mass transfer coefficients to be about 30 times higher in the water film than in the crust.