According to critical state theory, a soil will approach a critical void ratio during shear such that loose soils contract and dense soils dilate. Theory indicates that failing soils must be loose to generate the pore pressures needed for the mobilization of debris flows. Previously published results from large-scale experiments have also suggested that soils must be initially loose to fail as debris flows. In this contribution, this mechanism for soil liquefaction is tested in the field through observations and geotechnical analysis of soils that failed during a large storm in central California. Surprisingly, we find that the debris flows mobilized from soils that were initially dense. In addition, we find that the potential for debris flow mobilization was strongly linked to the fines/sand ratio. We present results from a numerical model that indicate that, as dilational soils approach the critical void ratio, the arresting effect of negative pore pressures generated by dilation is greatly reduced, leading to a rapid increase in basal pore pressure and rapid downslope acceleration. In addition, the model results suggest that the downslope displacement required to reach the critical state porosity in a dilative soil will be on the order of 0.1 to 1 m. Because the rate of the approach to critical state is fundamentally a function of the hydraulic conductivity of the soil, sandy soils will approach critical state much more rapidly than clay-rich soils.