While the importance of preferential flow in soils has been widely documented, it is not clear whether these pathways are continuous over long slope distances nor are the temporal aspects of preferential flow fully understood. Herein, concepts developed from numerous field-based investigations in Japan are summarized in the context of hillslope stormflow generation processes. In most forest hillslopes, individual soil macropores are relatively short (i.e., < 0.5 m); nevertheless different types of macropores have a tendency to self organize into larger preferential flow systems as sites become wetter. Staining tests show clear evidence of interconnected macropore flow segments including: flow within decayed root channels and subsurface erosion cavities; flow in small depressions of bedrock; fracture flow in bedrock; exchange between macropores and mesopores; and flow at the organic horizon/mineral soil interface and in buried pockets of organic material and loose soil. The spatially variable and non-linear preferential flow responses observed at the Japan field sites as well as other sites are attributed to discrete segments of macropores connecting at various nodes within the regolith. Each node is activated by local soil water conditions and is strongly influenced by soil depth, permeability, pore size, organic matter distribution, surface and substrate topography, and possibly momentum dissipation. Thus, preferential flow contributes increasingly to storm runoff in headwater catchments during wetter conditions. In contrast, Hortonian overland flow, which may be significant when measured at smaller scales and during relatively dry antecedent conditions, becomes a less important component of storm runoff as sites become wetter. Overland flow will often infiltrate into the soil as it progresses downslope, likely influenced by preferential flow paths. The interactions of spatially and temporally variable Hortonian overland flow and subsurface flow (including preferential flow) are presented in a conceptual model of stormflow generation processes.