Keywords: submarine landslides meteo-tsunami earthquakes and landslides probabilistic hazard assessment Tsunami hazard is a very low-probability, but potentially high-risk natural hazard, posing unique challenges to scientists and policy makers trying to mitigate its impacts. These challenges are illustrated in this assessment of tsunami hazard to the U.S. Atlantic margin. Seismic activity along the U.S. Atlantic margin in general is low, and confirmed paleo-tsunami deposits have not yet been found, suggesting a very low rate of hazard. However, the devastating 1929 Grand Banks tsunami along the Atlantic margin of Canada shows that these events continue to occur. Densely populated areas, extensive industrial and port facilities, and the presence of ten nuclear power plants along the coast, make this region highly vulnerable to flooding by tsunamis and therefore even lowprobability events need to be evaluated. We can presently draw several tentative conclusions regarding tsunami hazard to the U.S. Atlantic coast. Landslide tsunamis likely constitute the biggest tsunami hazard to the coast. Only a small number of landslides have so far been dated and they are generally older than 10,000 years. The geographical distribution of landslides along the margin is expected to be uneven and to depend on the distribution of seismic activity along the margin and on the geographical distribution of Pleistocene sediment. We do not see evidence that gas hydrate dissociation contributes to the generation of landslides along the U.S. Atlantic margin. Analysis of landslide statistics along the fluvial and glacial portions of the margin indicate that most of the landslides are translational, were probably initiated by seismic acceleration, and failed as aggregate slope failures. How tsunamis are generated from aggregate landslides remains however, unclear. Estimates of the recurrence interval of earthquakes along the continental slope may provide maximum estimates for the recurrence interval of landslide along the margin. Tsunamis caused by atmospheric disturbances and by coastal earthquakes may be more frequent than those generated by landslides, but their amplitudes are probably smaller. Among the possible far-field earthquake sources, only earthquakes located within the Gulf of Cadiz or west of the Tore-Madeira Rise are likely to affect the U.S. coast. It is questionable whether earthquakes on the Puerto Rico Trench are capable of producing a large enough tsunami that will affect the U.S. Atlantic coast. More information is needed to evaluate the seismic potential of the northern Cuba fold-and-thrust belt. The hazard from a volcano flank collapse in the Canary Islands is likely smaller than originally stated, and there is not enough information to evaluate the magnitude and frequency of flank collapse from the Azores Islands. Both deterministic and probabilistic methods to evaluate the tsunami hazard from the margin are available for application to the Atlantic margin, but their implementation requires more information than is c...