ABSTRACTIn measured sequences of limestone‐ and greywacke‐turbidites the bed‐thickness is found to vary proportionally with the fall velocity of the maximum grain size, found at the base of the bed. A simple theoretical model, based on the decay of isotropic turbulence, suggests that bed‐thickness should be a function not only of this fall velocity, but also of bottom slope, flow depth and the concentration and grain‐size distribution of the sediment in the turbidity current. The field data do show some influence of these additional factors. Nevertheless, for many natural sequences of turbidites the flows must have carried very poorly sorted sediments and the inferred flow volumes and densities must cluster tightly about modal values. Thus, grain size remains the primary variable and the modal regression curve of bed‐thickness on maximum grain size is well defined and resembles a fall‐velocity curve. Relatively steep basin floors near to source can, theoretically, cause these modal regressions for distal and very proximal parts of a turbidite to diverge, introducing a crudely parabolic appearance in the form of the total regression curve. The form of this parabolic curve predicts the deposition of thin but relatively coarse proximal beds. Such beds do occur. They are different from the thin, but relatively fine, proximal beds that have been interpreted as the result of a fractionation of a turbidity current during levee‐forming processes.