Communities in lowlands near volcanoes are vulnerable to significant volcanic flow hazards in addition to those associated directly with eruptions. The largest such risk is from debris flows beginning as volcanic landslides, with the potential to travel over 100 kilometers. Stratovolcanic edifices commonly are hydrothermal aquifers composed of unstable, altered rock forming steep slopes at high altitudes, and the terrain surrounding them is commonly mantled by readily mobilized, weathered airfall and ashflow deposits. We propose that volcano hazard assessments integrate the potential for unanticipated debris flows with, at active volcanoes, the greater but more predictable potential of magmatically triggered flows. This proposal reinforces the already powerful arguments for minimizing populations in potential flow pathways below both active and selected inactive volcanoes. It also addresses the potential for volcano flank collapse to occur with instability early in a magmatic episode, as well as the 'false-alarm problem'-the difficulty in evacuating the potential paths of these large mobile flows. Debris flows that transform from volcanic landslides, characterized by cohesive (muddy) deposits, create risk comparable to that of their syneruptive counterparts of snow and ice-melt origin, which yield noncohesive (granular) deposits, because: (1) Volcano collapses and the failures of airfall- and ashflow-mantled slopes commonly yield highly mobile debris flows as well as debris avalanches with limited runout potential. Runout potential of debris flows may increase several fold as their volumes enlarge beyond volcanoes through bulking (entrainment) of sediment. Through this mechanism, the runouts of even relatively small collapses at Cascade Range volcanoes, in the range of 0.1 to 0.2 cubic kilometers, can extend to populated lowlands. (2) Collapse is caused by a variety of triggers: tectonic and volcanic earthquakes, gravitational failure, hydrovolcanism, and precipitation, as well as magmatic activity and eruptions. (3) Risk of collapse begins with initial magmatic activity and increases as intrusion proceeds. An archetypal debris flow from volcanic terrain occurred in Colombia with a tectonic earthquake (M 6.4) in 1994. The Rio Piez conveyed a catastrophic wave of debris flow over 100 kilometers, coalesced from multiple slides of surflcial material weakened both by weathering and by hydrothermal alteration in a large strato- volcano. Similar seismogenic flows occurred in Mexico in 1920 (M -6.5), Chile in 1960 (M 9.2), and Ecuador in 1987 (M 6.1 and 6.9). Velocities of wave fronts in two examples were 60 to 90 km/hr (17-25 meters per second) over the initial 30 kilometers. Volcano flank and sector collapses may produce untransformed debris avalanches, as occurred initially at Mount St. Helens in 1980. However, at least as common is direct transformation of the failed mass to a debris flow. At two other volcanoes in the Cascade Range-- Mount Rainier and Mount Baker--rapid transformation and high mobility were typical of most of at least 15 Holocene flows. This danger exists downstream from many stratovolcanoes worldwide; the population at risk is near 150,000 and increasing at Mount Rainier. The first step in preventing future catastrophes is documenting past flows. Deposits of some debris flows, however, can be mistaken for those of less-mobile debris avalanches on the basis of mounds formed by buoyed megaclasts. Megaclasts may record only the proximal phase of a debris flow that began as a debris avalanche. Runout may have extended much farther, and thus furore flow mobility may be underestimated. Processes and behaviors of megaclast-bearing paleoflows are best inferred from the intermegaclast matrix. Mitigation strategy can respond to volcanic flows regardless of type and trigger by: (1) Avoidance: Limit settlement in flow pathways to numbers that can be evacuated after event warnings (flow is occurring). (2) Instrumental even