ABSTRACTMixing of fluids at fracture intersections was examined using both a series of plexiglass models and a two‐dimensional, finite‐element, discrete fracture model. The physical laboratory models included 12 models having two continuous, fully intersecting fractures with different intersection angles and apertures, a single model consisting of a single continuous fracture offsetting a second fracture, and a fracture system model consisting of parallel fractures in two intersecting sets. The plexiglass model results indicated essentially no mixing occurred in the fully intersecting fracture models when the apertures were equal. Mixing was found to be dependent only upon the relative size of the inlet and outlet fractures even with multiple intersections.For transport of a conservative solute in a discontinuous, random, discrete fracture system, the numerical model used the mixing algorithm for fracture intersections, developed from the physical model study. At each four‐way intersection, a novel approach was used to uncouple and recouple the nodal points to ensure the proper assignment of concentrations to each fracture element. Using the laboratory‐determined mixing algorithm, the numerical model demonstrated that more longitudinal and less lateral dispersion takes place than when complete mixing at fracture intersections is assumed. In addition, more longitudinal transport takes place in discontinuous than in continuous fracture systems. These findings indicate that contaminants migrating through fractured media, where the fracture walls are not in contact, will not be dispersed and diluted to the extent that previous numerical models have predicted; hence, the contaminant will be discharged to the biosphere in much greater concentration than e