The subgenus Lythrurus Jordan is diagnosed and its nomenclature discussed. Closely related to the atherinoides series of the subgenus Notropis Ralinesque, Lythrurus is composed of four species groups distinguished on the basis of tuberculation, vertebral number, and pigmentation. These groups are as follows: (1) the fumeus complex, includes only Notropis fumeus Evermann; (2) the ardens complex , includes N. ardens (Cope) and N. lirus (Jordan); (3) the umbratilis complex, includes N. umbratilis (Girard) and subspecies; and (4) the roseipinnis complex, which is revised herein. Members of the roseipinnis complex are distributed primarily in streams of the eastern Gulf Coast. All forms are closely related and distinguished chiefly on measurements and pigmentation, especially that of the fins. N. b. bellus (Hay) shows little geographic variation and is distributed throughout most of the Mobile Bay basin. It is absent above the Fall Line in the Coosa River system. Above the Fall Line in the Black Warrior River system it is replaced by a strongly differentiated new subspecies, N. b. alegnotus. These two forms appear to have interbred and formed intergrade populations around Tuscaloosa, Alabama. N. roseipinnis (Hay) is distributed in Gulf Coastal streams from Mobile Bay west through Lake Pontchartrain and in the Bayou Pierre, Big Black, and Yazoo drainages of the Mississippi Valley. This species exhibits striking geographic variation, much of an east-west clinal nature. Though they have not been collected together, N. b. bellus and N. roseipinnis live in close proximity in the lower Mobile Bay drainage. Here they are strongly differentiated, but more westerly populations of N. roselpinnis approach N. b. bellus in many features. N. atrapiculus, a new form previously confused with N. b. bellus and N. roseipinnis, is distributed in streams along the eastern Gulf Coast from the Escambia to the Apalachicola drainage. It is allopatric from its two close relatives and is intermediate or interjacent between them in most differentiating characters. Because of its intermediacy and the resulting weak differentiation, its proper taxonomic status is debatable. As nothing suggests whether it is more closely related to N. b. bellus or to N. roseipinnis, atrapiculus is accorded specific rank.