The R7 photoreceptor neuron projections form a retinotopic map in the medulla of the Drosophila optic lobe. The more inner photoreceptors mutation, an allele of gap1, results in the differentiation of excess R7s in the eye, whose axons invade the brain and establish functional connections. We have used this hyperinnervation phenotype to explore the roles of photoreceptor-target regulation, competitive interactions, and chemoaffinity in map formation. We show that the extra axons are supported in a wild-type brain, with all R7s from a single ommatidium sharing a single termination site, and thus there is no evidence that the target regulates the size of the presynaptic population. In mosaic eyes, in which ommatidia containing extra R7s are surrounded by ommatidia lacking all R7 cells, R7 axons still target to appropriate retinotopic locations in a largely empty R7 terminal field. Axons at the edges of the projection, however, send collaterals into vacant areas of the field, suggesting they are normally restrained to share single termination sites by competitive interactions. In contrast, no sprouts are seen when the vacant sites are juxtaposed with singly innervated sites. In the third instar, R7 and R8 axons transiently display halos of filopodia that overlap adjacent terminals and provide a means to assess occupancy at adjacent sites. Finally, in sine oculis larvae in which only a small number of ommatidia develop, the R7/R8 axons target to predicted dorsoventral portions of the medulla despite the absence of their neighbors, suggesting that position in the eye field determines their connectivity in the brain. We suggest that the mechanisms used to set up this insect map are formally similar to strategies used by vertebrates. The availability of a genetic model for these events should facilitate studies aimed at understanding the molecular bases of retinotopic map development.