A global data base consisting of radiocarbon-controlled relative sea level (RSL) histories from 392 different geographic locations has been compiled for use in analyses of the glacial isostatic adjustment process. These data may be employed to constrain both the spatiotemporal characteristics of the last deglaciation event of the current ice age and the rheological structure of the planetary interior. In a previous paper in this series (Tushingham and Peltier, 1991), we employed the RSL data in this set from sites that were once ice covered to refine the a priori model of deglaciation that had been delivered by past research in the area of Pleistocene geomorphology (Peltier and Andrews, 1976; Wu and Peltier, 1983) to produce a new global model of this event that we have called ICE-3G. In the present paper, following a description of the full data base and of the methods used to compile it, we employ data from the sites that were not ice covered (200 time series) to verify the plausibility of this refined model of deglaciation and to test the extent to which the totality of the observations constrain the radial viscoelastic structure of the planet. The analysis of solutions to the forward problem presented here reinforce previous inferences of a preference for a thick lithosphere by RSL data from sites along the U.S. east coast and a preference by data from most near-field sites for an increase in viscosity across the 670-km discontinuity in the mantle by a factor of approximately 2, when the viscosity of the upper mantle is fixed to the value of 1021 Pa s originally deduced by Haskel (1935) on the basis of his analysis of the data from Fennoscandia. ¿ American Geophysical Union 1992