SEA-LEVEL variations, as recorded by the global network of tide gauges, represent a rich data set for studying a nide range of natural and anthropogenic phenomena, such as the sea-level rise induced by possible global warming, For this purpose, long-term sea-level trends must be corrected for the ’contaminating’ effects of continuing glacial isostatic adjustment(1-5) (GIA). The numerical correction procedure has, for sites on the east coast of North America, yielded a set of highly anomalous sea-level rates characterized by systematic geographical trends(2,4,5). We demonstrate that these trends are a consequence of inadequacies in the previous ’standard’ numerical prediction for GIA, In particular, we find that the well-known trends in the GIA-corrected tide gauge rates are eliminated if the lower-mantle viscosity of the Earth model used in the GIA prediction is increased. This result obviates the need to explain the anomalous trend as a manifestation of Gulf Stream ocean circulation(4) or neotectonic processes(2).