The relationships of extant and extinct lineages of Adephaga were analysed formally for the first time. Emphasis is placed on the aquatic and semiaquatic groups and their evolution in the Mesozoic. †Triadogyrus and †Mesodineutus belong to Gyrinidae, the sister group of the remaining families. †Triaplidae are the sister group of the following groups (Haliplidae, Geadephaga, Dytiscoidea incl. †Liadytidae, †Parahygrobiidae and †Coptoclavidae [major part]). The lack of a ventral procoxal joint and a very short prosternal process are plesiomorphies of †Triaplidae. †Coptoclavidae and †Timarchopsinae are paraphyletic. †Timarchopsis is placed in a geadephagan clade. In contrast to other coptoclavids, its metathorax is close to the condition found in Haliplidae, with a complete transverse ridge and coxae with large plates and free mesal walls. †Coptoclavidae s.str., i.e. excl. †Timarchopsis, is a dytiscoid subgroup. The mesal metacoxal walls are fused, the coxal plates are reduced, and the transverse ridge is absent. †Stygeonectes belongs to this dytiscoid coptoclavid unit and is therefore misplaced in †Timarchopsinae. †Liadytidae belongs to a dytiscoid subgroup, which also comprises the extant families Aspidytidae, Amphizoidae, Hygrobiidae and Dytiscidae. †Parahygrobia is the sister group of Hygrobiidae. The larvae are characterized by a broad gula, the absence of the lacinia, retractile maxillary bases and very long urogomphi set with long setae. †Liadytiscinae is the sister group of extant Dytiscidae. There is no support for a clade †Eodromeinae and for Trachypachidae incl. †Eodromeinae. †Fortiseode is nested within Carabidae. The exclusion of fossil taxa has no effect on the branching pattern. The evolution of Adephaga in the Mesozoic is discussed. Possible reasons for the extinction of †Coptoclavidae are the rise of teleost fish and the competition of Gyrinidae and Dytiscidae, which possess efficient defensive glands and larval mandibular sucking channels.