Late Holocene relative sea-level (RSL) reconstructions can be used to estimate rates of land-level (subsidence or uplift) change and therefore to modify global sea-level projections for regional conditions. These reconstructions also provide the long-term benchmark against which modern trends are compared and an opportunity to understand the response of sea level to past climate variability. To address a spatial absence of late Holocene data in Florida and Georgia, we reconstructed ~ 1.3 m of RSL rise in northeastern Florida (USA) during the past ~ 2600 years using plant remains and foraminifera in a dated core of high salt-marsh sediment. The reconstruction was fused with tide-gauge data from nearby Fernandina Beach, which measured 1.91 ± 0.26 mm/year of RSL rise since 1900 CE. The average rate of RSL rise prior to 1800 CE was 0.41 ± 0.08 mm/year. Assuming negligible change in global mean sea level from meltwater input/removal and thermal expansion/contraction, this sea-level history approximates net land-level (subsidence and geoid) change, principally from glacio-isostatic adjustment. Historic rates of rise commenced at 1850–1890 CE and it is virtually certain (P = 0.99) that the average rate of 20th century RSL rise in northeastern Florida was faster than during any of the preceding 26 centuries. The linearity of RSL rise in Florida is in contrast to the variability reconstructed at sites further north on the U.S. Atlantic coast and may suggest a role for ocean dynamic effects in explaining these more variable RSL reconstructions. Comparison of the difference between reconstructed rates of late Holocene RSL rise and historic trends measured by tide gauges indicates that 20th century sea-level trends along the U.S. Atlantic coast were not dominated by the characteristic spatial fingerprint of melting of the Greenland Ice Sheet.