Subaqueous paleoseismology is more than 30 years old, but is just beginning to demonstrate its potential for the future. Long paleoseismic records are our best hope of understanding long-term patterns of energy release. However, to be an effective tool, turbidite paleoseismology requires careful consideration of site context, temporal interval of interest, sediment supply, flow pathways, and flow dynamics. In Cascadia, numerous channel systems exist and cover a range of time intervals. During high stands, many of these systems are relict, with limited terrigenous sediment supply, while some are indirectly recharged through cross-shelf transport from river systems. In relict systems, Holocene paleoseismic records commonly depend on recycled materials from failure of local slopes. Local failures may serve to supply sediment at any point along a canyon system or basin under expected shaking levels, with or without recent sediment recharge. Recharge by active terrigenous sedimentation is apparently not required in Cascadia, Sumatra, or the Caribbean, where site locations without this recharge possibility have excellent records correlable to other sites. However, oversupply of relict sediment may dwindle over time leaving a narrow window of time for paleoseismic records. By comparison to Pleistocene fan-building currents in Cascadia, Holocene currents are weak, rendering most areas of fan systems inactive. Pleistocene channels are crosscut by active Holocene incisions and levees, restricting their role as depocenters. In the main channels, the most recent currents are largely confined closely within their levees, showing evidence of overbank deposition extending 5-8 km from the channel, and consistent with turbidity current heights less than 150 m. In Sumatra, no modern recharge of slope basins is possible due to the isolation of the outer forearc slope. There, canyons and channels are few, and most basin fill is derived from failure of local slopes, carrying recycled material and foram sand from shallower levels. Transport distances are short in unchannelized systems, thus sampling location near local slopes is critical to obtaining a paleoseismic record. The growing family of long paleoseismic records globally will soon make it possible to address long-term cycling and clustering of Great earthquakes.