An assay has been developed to test the hypothesis that neurons from a limited area of the retina will adhere preferentially to that part of the optic tectum near their normal synaptic termini. The method measures the adherence of (32)P-labeled cell bodies from either the dorsal or ventral half of the neural retina of chick embryos to dorsal and ventral tectal halves. When a labeled, single-cell suspension is prepared from dorsal half-retina, more cells bind to the ventral half of the tectum. When the labeled cells are from ventral half-retina, more bind to dorsal half-tecta. This preferential adhesion mimics the retinotectal projection found in vivo. Dorsal retinal cells show this preference shortly after dissociation with crude trypsin, and maintain it for at least 9 hr. Ventral retinal cells, however, require incubation in nutrient medium after trypsinization in order to display this selectivity. Comparable results are obtained when the cell suspension is prepared from pigmented retina. The data support an interpretation of neuronal specificity dependent on cell-surface interactions and demonstrate a clear correlation between selective adhesion and biological function.