Watersheds are self-organized natural entities, consisting of interacting, inter-dependent parts, e.g., climate, soils, vegetation, topography (hillslopes), and the river network. They are characterized by strong (usually non-linear) interactions between the parts, including significant thresholds, and complex feedback loops that make it difficult to distinguish between cause and effect. In fact, the watershed is a metaphor for integration. Yet, traditional reductionist approaches used in hydrology tend to split the watersheds into smaller and smaller isolated parts, assuming that the parts can be added up to give the behaviour of the whole. One can cite many examples to demonstrate that the reductionist paradigm has led us to a theoretical impasse. Models of considerable sophistication have been developed, that are yet unable to make predictions in ungauged watershed without local tuning, i.e., calibration. Current theories of water movement in soils (e.g. Darcy’s law) are at odds with observations of preferential flow in networks of pipes and fractures. The old water paradox—the fact that watersheds store water for considerable periods of time and then release it promptly during rain events cannot be explained by existing theories. Current observations are too sparse or inappropriate, and methods of data analysis inadequate, for the elucidation of underlying patterns, especially patterns of emergent behavior that may be caused by threshold non-linearities. Apart from the almost ritualistic recourse to mass and momentum balance principles, we have not yet discovered hydrological laws or principles that underpin the interactions and/or feedbacks between parts of a watershed that are useful for predictions. It is abundantly clear that the current hydrological perspective is too narrow, and current datasets and methods of data analysis are too limited. We must therefore broaden our hydrologic perspective by forming new, stronger links with disciplines such as climatology, geomorphology, ecology, geophysics etc. and by embracing the new datasets, new methods of data analysis and new concepts that such links will make possible. It is through such multi-disciplinary interactions that we can ever achieve the much hoped-for paradigm change in watershed hydrology.