Extreme uplift rates and sea level changes in southern Alaska have been documented by Global Positioning System (GPS) surveys, tide gauge measurements and studies of raised shorelines. The movements detected in a network of 45 GPS survey points describe a broad pattern of rapid regional uplift. The majority of the study area is uplifting at a rate faster than 10 mm yr-1, with several sites uplifting more rapidly than 25 mm yr-1. New tide gauge data presented here consist of repeat occupations of 18 temporary gauge sites. Sea level rates at these sites agree with similar measurements in southern Alaska taken ~50 yr earlier, and also with the pattern of uplift derived from the GPS data. Raised shoreline studies at 14 sites document total sea level change, with a maximum change in sea level of -5.7 m found in upper Lynn Canal. The start of the ongoing uplift episode that raised these shorelines has been dated with dendrochronology and found to be coincident with the start of the collapse of the Glacier Bay Icefield, at ca. 1750 AD. The pattern of total sea level change is in general agreement with uplift-rate measurements, with greater sea level change found at the sites closest to the peak uplift rates in upper Glacier Bay. We use a viscoelastic earth model subjected to an ice load history built upon observations of glacial change to predict uplift rates at the tide gauge and GPS sites as well as the total uplift at the raised shoreline sites. Our modelling exercises are limited to an ice load model based on independent studies of the region's glacial history over the past 1.7 kyr, to evaluate whether the uplift observations can be explained by simple earth models subjected to this load history. Two-layer earth models, consisting of an elastic crust and a low-viscosity upper mantle half-space, can be adjusted to fit either the raised shoreline data or the combined GPS and tide gauge uplift-rate data, but cannot fit all the data with a single set of earth model parameters. However, all three data sets are consistent with an approximated three-layer earth model. The combined model is constrained by a total of 77 uplift measurements, which at the 95 per cent confidence level require a low-viscosity asthenosphere [ηA= (1.4 +/- 0.3) × 1019 Pa s and thickness 110+20-15 km] beneath a 50+30-25 km thick elastic lithosphere and overlying an upper mantle half-space with a viscosity of 4 × 1020 Pa s. This earth model achieves a low degree of misfit with the observations (reduced chi-square value of χ2ν= 2.5), suggesting that glacial isostatic rebound associated with post-Little Ice Age melting can entirely account for the rapid uplift of southern Alaska over the last ~250 yr.