The initial explosions at Mount St. Helens, Washington, on the moring of 18 May 1980 developed into a huge pyroclastic surge that generated catastrophic floods off the east and west flanks of the volcano. Near-source surge deposits on the east and west were lithic, sorted, lacking in accretionary lapilli and vesiculated ash, not plastered against upright obstacles, and hot enough to char wood - all attributes of dry pyroclastic surge. Material deposited at the surge base on steep slopes near the volcano transformed into high-concentration lithic pyroclastic flows whose deposits contain charred wood and other features indicating that these flows were hot and dry. Stratigraphy shows that even the tail of the surge had passed the east and west volcano flanks before the geomorphically distinct floods (lahars) arrived. This field evidence undermines hypotheses that the turbulent surge was itself wet and that its heavy components segregated out to transform directly into lahars. Nor is there evidence that meters-thick snow-slab avalanches intimately mixed with the surge to form the floods. The floods must have instead originated by swift snowmelt at the base of a hot and relatively dry turbulent surge. Impacting hot pyroclasts probably transferred downslope momentum to the snow surface and churned snow grains into the surge base. Melting snow and accumulating hot surge debris may have moved initially as thousands of small thin slushflows. As these flows removed the surface snow and pyroclasts, newly uncovered snow was partly melted by the turbulent surge base; this and accumulating hot surge debris in turn began flowing, a self-sustaining process feeding the initial flows. The flows thus grew swiftly over tens of seconds and united downslope into great slushy ejecta-laden sheetfloods. Gravity accelerated the floods to more than 100 km/h as they swept down and off the volcano flanks while the snow component melted to form great debris-rich floods (lahars) channeled into valleys. ?? 1989 Springer-Verlag.