U.S. Atlantic margin cold seeps database from multibeam water column imagery, 2011-2016: South Atlantic Bight to Georges Bank. This release includes three data sets that provide geographic positions and water depths for cold seeps identified on the U.S. Atlantic margin between ~50 and ~2650 m water depth from the Blake Ridge on the south to the most seaward part of the southern New England margin on the north. The data were collected on cruises completed by NOAA’s R/V Okeanos Explorer (operated by the Office of Ocean Exploration and Research) or academic research vessels (e.g., R/V Atlantis, R/V Armstrong, and R/V Sikuliaq) between 2011 and 2016. The raw data are available from NCEI’s water column sonar data portal (https://www.ncei.noaa.gov/products/water-column-sonar-data). Details about locating seeps in the raw data are provided in Skarke et al. (2014) and Ruppel et al. (2024). The first data set in this release contains 2052 raw seep identifications, with many places sampled multiple times during subsequent passes on the same cruise or during subsequent cruises. This raw data set therefore includes duplicates, but does report the cruise, trackline, date, and time for the identified water column plume, along with a quality factor and other notes. The second data set, which is the one most useful for plotting unique seep locations, has 1139 seeps culled from the raw dataset by using the density-based clustering with noise algorithms within QGIS software to remove locations that were within 40 m of each other. The third data set applies the same QGIS cluster analysis to the second data set using a search radius of 400 m, yielding 47 seep clusters (5 to 138 seeps each) and 227 identifications outside of clusters. The resulting 274 locations represent the unique seep fields or seep sites along the margin. Details about the analyses are provided in Ruppel et al. (2024). The initial seep identifications used in the raw data set were made by analyzing water column data (WCD) collected by multibeam sonar systems, particularly 30 kHz systems like the EM302. A small amount of the analyzed multibeam data WCD came from EM70 systems.