Two sets of recent findings draw our attention to questions concerning the origin of ion channels. First, there is sequence similarity among five classes of channels: voltage-gated channels, a putative Ca(2+)-activated K+ channel, cyclic nucleotide-gated cation channels, a putative Ca2+ channel for phosphoinositide-mediated Ca2+ entry, and a plant K+ channel/transporter. Like voltage-gated K+ channels, the most recently identified members of the superfamily share the basic design of one set of six potential membrane-spanning segments plus the H5 sequence; as such, they may resemble more closely the ancestral channel, which is likely to predate the separation of the animal and plant kingdoms. Second, several members of the ABC superfamily function as ion channels, even though they were previously known as transporters or enzymes. Did some ancestral enzymes subsequently acquire channel/transporter function? Or could it be the other way around? Aside from evolutionary considerations, enzymes and ion channels can no longer be treated as separate and nonoverlapping groups of proteins. When one molecule exhibits both functions, there are interesting mechanistic questions: How might the enzyme activity such as ATP hydrolysis be coupled to activation/regulation of the intrinsic channel activity? How might interactions between the permeant ions and the channel pore in turn regulate the enzymatic function of the same molecule? It seems possible that the latter is an extension of the observed coupling between permeant ions and the gating machinery of an ion channel (Swenson and Armstrong, 1981). Finally, the potential cross-regulation between channel activity and enzyme activity within the same molecule offers many intriguing possibilities for the integration of different cellular functions.