Gap junctions (GJs) are composed of proteins from two distinct families. In vertebrates, GJs are composed of connexins; a connexin hexamer on one cell lines up with a hexamer on an apposing cell to form the intercellular channel. In invertebrates, GJs are composed of an unrelated protein family, the innexins. Different connexins have distinct properties that make them largely non-interchangeable in the animal. Innexins are also a large family with high sequence homology, and some functional differences have been reported. The biological implication of innexin differences, such as their ability to substitute for one another in the animal, has not been explored. Recently, we showed that GJ proteins are necessary for the development of normal neural transmission in the Drosophila visual system. Mutations in either of two Drosophila GJ genes (innexins), shakB and ogre, lead to a loss of transients in the electroretinogram (ERG), which is indicative of a failure of the lamina to respond to retinal cell depolarization. Ogre is required presynaptically and shakB(N) postsynaptically. Both act during development. Here we ask if innexins are interchangeable in their role of promoting normal neural development in flies. Specifically, we tested several innexins for their ability to rescue shakB(2) and ogre mutant ERGs and found that, by and large, innexins are not interchangeable. We mapped the protein regions required for this specificity by making molecular chimeras between shakB(N) and ogre and testing their ability to rescue both mutants. Each chimera rescued either shakB or ogre but never both. Sequences in the first half of each protein are necessary for functional specificity. Potentially crucial residues include a small number in the intracellular loop as well as a short stretch just N-terminal to the second transmembrane domain. Temporary GJs, possibly between the retina and lamina, may play a role in final target selection and/or chemical synapse formation in the Drosophila visual system. In that case, specificity in GJ formation or function could contribute, directly or indirectly, to chemical synaptic specificity by regulating which neurons couple and what signals they exchange. Cells may couple only if their innexins can mate with each other. The partially overlapping expression patterns of several innexins make this 'mix and match' model of GJ formation a possibility.