Determining the earthquake segmentation of plate-boundary transform faults remains a scientific challenge because paleoseismic data sets rarely resolve the end points of past ruptures. In this study, we test whether lacustrine paleoseismology can be used to assess rupture end points and the earthquake segmentation of the Alpine fault, one of the longest and fastest-slipping plate-boundary transform faults on Earth. Sediment cores from Lake Ellery record eight episodes of high-intensity shaking (modified Mercalli intensity [MM] IX) from Alpine fault earthquakes as event sequences of a turbidite produced by coseismic subaqueous mass wasting, overlain by deposits representing sediment flux from co- and postseismic landsliding in the fluvial catchment. Age-depth modeling constrains the timing of shaking events at a decadal resolution, facilitating correlation with two previously published lake records to reconstruct the spatial distribution of MM IX shaking along ∼150 km of the Alpine fault. When resolved with existing on- and near-fault paleoseismic records, the lake data set demonstrates that independent ruptures of the South Westland and Central segments occurred in A.D. 845–775 and A.D. 739–646, and A.D. 646–592 and A.D. 416–370, respectively. Lakes adjacent to the Alpine fault provide paleoseismic records with sufficient spatial and temporal resolution to define along-strike differences in the pattern of rupture capable of distinguishing rupture termination at a geometric segment boundary. This multilake study suggests that locating the end points of ruptures using lacustrine paleoseismology will be most applicable in midlatitude convergent plate-boundary settings where along-strike topography and Quaternary glaciation have resulted in the widespread distribution of suitable lakes.