This review provides a historical summary of the development of knowledge on preferential and non-equilibrium flows in soils, particularly in the period 1864 to 1984. It is pointed out that preferential flows were recognized long before the equilibrium concepts of Edgar Buckingham and Lorenzo A. Richards were developed and became the dominant underpinnings of soil physics in the 20th century, effectively in denial of all the evidence that the Buckingham-Richards theory was inadequate to deal with water flows in field soils. Summaries of evidence from soil microscopy, tracing and breakthrough curve experiments, infiltration and throughflow observations, and studies of natural pipes are presented. Approaches to modeling preferential flows at soil profile and hillslope responses are also reviewed, including viscous flow theory and particle tracking methods. Further research is required into the integration of viscosity and capillarity-dominated flows, into nonlaminar flows in larger pores, and into upscaling from core and profile experiments to larger hillslope scales. Also, more rigorous hypothesis testing is required in soil physics and hillslope hydrology using both flow and tracer data. This might lead to a new paradigm in representing the detail complexity of flow in soils in applications at larger scales.