Groundwater flow and solute transport are affected by the connectivity of high hydraulic conductivity (K) sediments. This research used numerical modeling and field experiments to assess and quantify connectivity and to compare alternative transport models and approaches for representing connected heterogeneity. Field work was conducted at the well-known Macrodispersion Experiment (MADE) site to investigate transport hypothesized to be controlled by a network of highly connected preferential flow paths (PFP). The research results in three self-contained, closely related, papers. The first paper evaluates the ability of different transport models to reproduce the transport behavior in a synthetic 2-D aquifer system with a high-K network embedded in a low-K matrix. Results confirm that the classical Fickian advection-dispersion model (ADM) is unable to effectively reproduce solute transport unless heterogeneity is explicitly considered. Conversely, two non-Fickian models (dual-domain mass transfer and continuous-time random walk) are able to accurately match the transport behavior using only effective parameters. However, the continuous time random walk model requires a calibrated transport velocity that is physically unrealistic. The second paper investigates flow and transport connectivity in a small block of the MADE site aquifer. K values estimated from grain size analysis of 19 cores are used to generate 3-D conditional realizations of the K field. Anomalous transport in the generated K fields is revealed by particle tracking simulations and significant connectivity is quantified by a variety of connectivity indicators. Particle paths geometry shows that flow and transport connectivity do not require fully percolating high-K clusters. The third paper presents the results of new tracer test. Breakthrough curves measured at the extraction well and at 14 multilevel sampling ports along the vertical extension of the MADE site aquifer clearly reveal the presence of a complex network of PFPs. Numerical modeling based on experimental data shows that the dual-domain mass transfer model successfully captures the characteristics of the integrated breakthrough curve at the extraction well, but it is ineffective in reproducing the concentrations observed at the multilevel sampling locations, indicating that a high-resolution characterization of the aquifer heterogeneity would be needed to fully capture 3D transport details.