The processes occurring in the basal region of concentrated pyroclastic density currents (PDCs) influence the mobility, runout distance, and damage potential of a current, but directly observing these processes is extremely difficult. Instead, we must investigate the deposits to glean information regarding the conditions of sediment transport and deposition. The PDC deposits of the May 18, 1980 eruption of Mount St. Helens (WA, USA) contain sedimentary structures consisting of bed material reworked into undulose structures and recumbent flame structures. The structures vary over two orders of magnitude in size with lengths ranging from 8 cm to 18 m and heights ranging from 4 cm to 1.8 m. Despite the large range in sizes, the structures remain self-similar in form, suggesting a common mechanism for formation. The structures are interpreted as the record of granular shear instabilities, similar to Kelvin-Helmholtz instabilities, formed at the interface between a shearing, high-concentration flow and the substrate in the moments just prior to deposition. The morphology of the structures suggests that the basal region of PDCs must be both highly concentrated and also highly mobile in the moments before final deposition, likely a result of elevated pore fluid pressures. We use a modified instability growth criterion to estimate PDC flow velocities at the time of formation; for the Mount St. Helens PDCs, the velocity estimates range from 0.2 to 7.5 m s−1 with larger structures requiring higher flow velocities. Combining the velocity estimates with the dimensions of the structures suggests deposition rates of 4 to 32 cm s−1. Such high deposition rates indicate that the deposits likely accumulated in a stepwise manner, rather than either progressively or en masse. The structures suggest that sections of the deposit accumulated during punctuated periods of high deposition lasting at most a few seconds followed by periods of bypassing (i.e., non-deposition) or erosion lasting minutes to tens of minutes. Our findings motivate continued experimental and numerical work to understand how the formation of recumbent flame (and similar) structures affects subsequent flow behavior in terms of runout distance and hazard potential.