In this paper we adopt a quantitative biostratigraphic approach to establish a 1000-year-long coastal record of megathrust earthquake and tsunami occurrence in south central Chile. Our investigations focus on a site in the centre of the rupture segment of the largest instrumentally recorded earthquake, the AD 1960 magnitude 9.5 Chile earthquake. At Chucalén coseismic subsidence in 1960 is recorded in the lithostratigraphy and biostratigraphy of coastal marshes, with peat overlain by minerogenic sediment and changes in the assemblages of diatoms (unicellular algae) indicating an abrupt increase in relative sea level. In addition to the 1960 earthquake, the stratigraphy at Chucalén records three earlier earthquakes, the historically documented earthquake of 1575 and two prehistoric earthquakes, radiocarbon dated to AD 1270–1450 and 1070–1220. Laterally extensive sand sheets containing marine or brackish diatom assemblages suggest tsunami deposition associated with at least two of the three pre-1960 earthquakes. The record presented here suggests a longer earthquake recurrence interval, averaging 270 years, than the historical recurrence interval, which averages 128 years. The lack of geologic evidence at Chucalén of two historically documented earthquakes, in 1737 and 1837, supports the previously suggested hypothesis of variability in historical earthquake characteristics. Our estimates of coseismic land-level change for the four earthquakes range from meter-scale subsidence to no subsidence or slight uplift, suggesting earthquakes completing each ∼270 year cycle may not share a common, characteristic slip distribution. The presence of buried soils at elevations below their modern equivalents implies net relative sea-level rise over the course of the Chucalén paleoseismic record, in contrast to relative sea-level fall over preceding millennia inferred from sites on the mainland. Sea-level rise may contribute to the preservation of evidence for multiple earthquakes during the last millennium, while net relative sea-level fall over the last 2000–5000 years may explain the lack of evidence for older earthquakes.