We use a database of CRRES in situ observations of plasmapause crossings to build empirical models of the plasmapause location as a function of geomagnetic indices. Previous models used the maximum value in Kp during the hours to days leading up to the plasmapause crossing. We find that a recent maximum in AE or minimum in Dst provides a better model of the plasmapause radius than does maximum Kp. AE and Dst measure specific current systems (the auroral electrojet and ring current, respectively). The AE model suggests that substorms may be involved in the erosion of the plasmapause. The Dst model suggests that ring current may be formed by the same electric field that erodes the plasmapause. In more complex models, Kp and AE can describe local time structure in the plasmapause, with the furthest plasmapause before dawn at quiet times, premidnight at active times.