A principle question concerning storm time convection is. What physical process or measurable parameter controls the location of the equatorward edge of the large-scale convection zone expansion during magnetic storms and hence the plasmapause location? Experimental data on convection and particle precipitation consistently show that in the evening to midnight local time sector, the inner boundary of the high-latitude westward ion convection band colocates with the equatorward boundary of soft electron precipitation (SEB). Low-energy electron precipitation is usually absent, or very weak, above the lower-latitude band of the disturbed-time double-peaked convection pattern. It follows that large-scale convection streamlines carrying plasma sheet particles do not enter the polarization jet band which lies on the opposite (inner) side of the Alfven layer, which limits the inward expansion of hot plasma and convection by polarizing the edge of the inner magnetosphere plasma population. We conclude that the SEB (measured and/or modeled) can be used in plasmasphere density models as a substitute for the convection boundary. A time-dependent convection-driven plasmaspheric density model (CDPDM) is introduced to describe plasmaspheric thermal density profiles. The CDPDM is based on the convection drift and refilling rate prehistory calculated for a particular plasma flux tube, and its most important ingredient is a realistic convection model for disturbed times. Sharp density gradients (plasmapauses) on the radial profiles are indicators of preceding convection boundary locations outside of which the thermal plasma content was lost. We compare the predictions of the model with storm time ionospheric observations with the Millstone Hill radar and conclude that the CDPDM can be used to predict the locations of plasma density radial gradients, including the plasmapause.¿ 1997 American Geophysical Union