Spatial and temporal variations in sea-level are produced by the melting of the Late Pleistocene ice and by the Earth's response to the redistribution in surface loads. By examining different parts of the sea-level curves of the past 20 000 yr from geographically widely distributed regions it becomes possible to constrain models of the melting history of the ice sheets and of the Earth's rheology. Observations from sites away from the former Arctic ice sheets, such as the Australian and South Pacific region, are particularly important for constraining the total meltwater volumes added into the oceans in the past 20 000 yr and the rates at which this occurred. These observations indicate that the Antarctic ice sheets provided a significant contribution to the sea-level rise at a rate that was approximately synchronous with the melting of the Laurentide ice sheet, except for the interval 9000–6000 yr ago when it may have lagged behind. Minor melting of the Antarctic ice sheet appears to have continued throughout the Late Holocene. Differential observations of the Late Holocene sea-level change recorded at sites in the same region are particularly useful for estimating parameters describing the Earth's non-elastic response to surface loading. The effective parameters used here are a lithospheric thickness, an upper mantle viscosity, and a lower mantle viscosity describing the response below 670 km depth. With the observations used here, it is not possible to separate the lithospheric thickness H from the upper mantle viscosity and the viscosity results are based on the assumption that 50 ≤H≤ 100 km. Neither do these observations provide a resolution of the depth dependence of viscosity in the upper mantle and the resulting estimates are effective parameters only. Differential sea-levels along continental margins and along the shores of large gulfs and bays constrain the effective upper mantle viscosity to be about (1–2) × 1020 Pa s−1 while differential values from islands of different sizes are suggestive of a somewhat lower value. The lower mantle (taken to be below 670 km depth) viscosity is about two orders of magnitude greater than this. The estimated ice and rheological models explain many of the Holocene sea-level observations throughout the Australian and southern Pacific region. The tilting of continental margins, as exemplified by observations of variable amplitudes of the Holocene high-stands and the variable times at which sea-levels first reached their present level along the north Queensland coast and Great Barrier Reef, is well represented by these models. Differential Holocene sea-levels observed along the narrow Spencers Gulf of South Australia are also well explained by the models and in neither case is it necessary to invoke tectonic motions. Predicted Late Holocene sea-levels at small-to medium-sized islands are characterized by small amplitude high-stands that reached their maximum values about 4000–2000 yr ago, consistent with observations from the Society, Cook and Tuamotu Islands.