Excitation of fly photoreceptor cells is initiated by photoisomerization of rhodopsin to the active form of metarhodopsin. Fly metarhodopsin is thermostable, does not bleach, and does not regenerate spontaneously to rhodopsin. For this reason, the activity of metarhodopsin must be stopped by an effective termination reaction. On the other hand, there is also a need to restore the inactivated photopigment to an excitable state in order to keep a sufficient number of photopigment molecules available for excitation. The following findings reveal how these demands are met. The photopigment undergoes rapid phosphorylation upon photoconversion of rhodopsin to metarhodopsin and an efficient Ca2+ dependent dephosphorylation upon regeneration of metarhodopsin to rhodopsin. Phosphorylation decreases the ability of metarhodopsin to activate the guanine nucleotide-binding protein. Binding of 49-kDa arrestin further quenches the activity of metarhodopsin and protects it from dephosphorylation. Light-dependent binding and release of 49-kDa arrestin from metarhodopsin- and rhodopsin-containing membranes, respectively, directs the dephosphorylation reaction toward rhodopsin. This ensures the return of phosphorylated metarhodopsin to the rhodopsin pool without initiating transduction in the dark. Assays of rhodopsin dephosphorylation in the Drosophila retinal degeneration C (rdgC) mutant, a mutant in a gene previously cloned and predicted to encode a serine/threonine protein phosphatase, reveal that phosphorylated rhodopsin is a major substrate for the rdgC phosphatase. We propose that mutations resulting in either a decrease or an improper regulation of rhodopsin phosphatase activity bring about degeneration of the fly photoreceptor cells.