This dataset consists of a national 30-meter raster which classifies pixels representing water from the National Land Cover Database 2016 (NLCD 2016) according to their estimated velocity regime. At the broadest classification, the data are mapped into two groups: "Lentic", representing water bodies such as lakes, ponds, oceans, and bays, and "Lotic", representing flowing water, such as streams, rivers, and canals. These are further broken into 12 subordinate classes ordered according to velocity. These include categories of retention time for water in lakes/ponds and flow velocity for streams/rivers. The classes are derived based on information from the NHD High- and Medium-Resolution databases. Of the 470.7 million water pixels in the NLCD, approximately 143 million are classified as ocean/bay/estuary (30%), 295 million as lakes/ponds/reservoirs (63%), and 32 million as streams/rivers/canals (7%). The dataset allows users to better characterize the nature of flowing and non-flowing water when using the NLCD.