Metabolism of cGMP is critically important for the functioning of phototransduction in the mammalian retina. In rod and cone photoreceptors, two types of antagonistic enzymes, guanylate cyclases and cGMP phosphodiesterases, carefully balance the available amount of the intracellular messenger. Guanylate cyclase produces cGMP and phosphodiesterase rapidly hydrolyzes cGMP upon bleaching of the photopigment. Regulation of their activity in light and dark, influence of Ca++, and feed-back mechanisms are currently under intense investigation. A molecular analysis on both the gene and protein levels will contribute significantly to our understanding of their respective roles in phototransduction. The two types of enzymes have been characterized molecularly to a very different extent. Rod phosphodiesterase was purified to homogeneity almost fifteen years ago, but photoreceptor guanylate cyclase has evaded all attempts for molecular characterization. Characterization of retinal guanylate cyclase cDNA(s), however, will most likely be achieved in the near future. Cone PDE was shown to be a distinct enzyme, different from, but related to, the rod enzyme. Molecular cloning has provided sequence information of two of the three subunits of rod PDE; the small inhibitory subunit has been expressed in bacterial expression vectors, giving us an elegant tool for exploring mechanisms of activation and inhibition. The gene encoding the alpha subunit was shown to be a member of a large gene family of cyclic nucleotide phosphodiesterases, present in many eucaryotes ranging from unicellular organisms (yeast) to mammals. While much has been achieved, many questions remain to be answered. The beta subunit of rod phosphodiesterase has evaded complete molecular characterization, and its origin (one gene and posttranslational modification of the gene product generating alpha and beta, alternative splicing, or two separate genes with distinct gene products) has not been elucidated. Mechanisms of interaction of subunits, activation and inhibition, the active site(s) of the enzyme are undefined. Virtually nothing is known about the molecular organization of the photoreceptor guanylate cyclase(s). Recent cloning of two apparently unrelated mammalian guanylate cyclases, however, containing a common homologous domain signals increasingly rapid progress in this field.