A thin sedimentary blanket, consisting mostly of subtidal, unconformity-bounded calcarenite units, was deposited in the small Novoli graben (Apulian foreland, southern Italy) in Pliocene2Pleistocene time. In a limited part of the study area the lower Pleistocene ‘‘Calcarenite di Gravina,'' forming the thicker part of this blanket, is crossed by continuous to discontinuous cylindrical pipes as much as 12 m high, most commonly consisting of stacked concaveupward laminae, locally grading upward into soft-sediment-deformation features and large dishes. The evidence favors an origin linked to upwelling of overpressured groundwater from a large karstic reservoir hosted in the Mesozoic carbonate rocks; the reservoir periodically developed a relatively high hydrostatic head due to Tertiary to Pleistocene cover acting as an aquitard or aquiclude. As a result, submarine springs were generated, the activity of which was primarily controlled by relative sea-level fluctuations. It is suggested that the pipes were located in those points where the hydrostatic pressure was sufficient to fluidize the overlying sediment and could be released without notably affecting the surrounding sediments. Some pipes cross calcarenitic infills of karstic sinkholes developed in the underlying units, whereas others follow the course of vertical to high-angle extensional synsedimentary tectonic fractures generated when the calcarenites were still in an unconsolidated to semiconsolidated state. The former relationships suggest that vertical routes of water upwelling during highstand of base level commonly coincided with axes of vadose solution during base-level lowstand; the latter suggest that opening of fractures enhanced the connection of the deep aquifer with the surface, hence intensifying water upwelling. We think that fluidization along the fractures was not hindered by the partially coherent state, and that pipes with a cylindrical geometry could form in spite of the planarity of the fractures. The formation of the pipes and their internal structure of stacked concave-upward laminae is thought to be consistent with a process of fluidization due to through-flowing waters. We believe that essential in this process is the role of upward-migrating transient water-filled cavities, akin to the voidage waves (Hassett's [1961a, 1961b] parvoids) experimentally reproduced by several authors in liquid fluidized beds, and regarded as true instability phenomena of a fluidized suspension occurring above minimum fluidization velocity. It is suggested that the process is akin to the production of the dish structure. It consists of the filling of transient, upward-migrating, water-filled cavities through steady fallout of particles from the cavity roof, their redeposition in a more consolidated state, and subsidence of the roof due to water seepage upward from the cavity. The process was accompanied by segregation of grains according to their size and density, as well by elutriation of finest particles, and led to a new pattern of sediment texture, packing, and fabric with respect to the surrounding calcarenites.