This dataset contains 7,206,623 tweet IDs of Twitter activity during all major U.S. hurricanes since Twitter's inception in 2006. The tweets were harvested by focusing on users mentioning the words "storm," "superstorm," and "hurricane" 1 month before landfall, during, and 1 month after dissipation. Data is collected for the following hurricanes: Arthur (2014), Florence (2018), Gustav (2008), Harvey (2017), Hermine (2016), Humberto (2007), Ike (2008), Irene (2011), Irma (2017), Isaac (2012), Jose (2017), Maria (2017), Matthew (2016), Michael (2018), Nate (2017), Patricia (2015), and Sandy (2012). Due to Twitter’s Developer Policy, only tweet IDs can be publicly shared for academic purposes, which is why the .csv file only contains tweet IDs. These can easily be "hydrated" using a hydrating tool such as Hydrator or Twarc . Citation: Tacheva, Zhasmina, 2019, "U.S. Hurricane Tweets (2006-2018)", https://doi.org/10.7910/DVN