In this review we have discussed the problem of deactivation at both the rhodopsin and G protein levels. Of particular interest is the novel observation that rhodopsin deactivation can be modulated by light. This modulation is likely to play an important role in light adaptation by reducing the gain of transduction. One interesting possibility is that this modulation involves the phosphorylation of an arrestin-like molecule, but this remains to be tested. One of the experimental advantages of Limulus photoreceptors is the large size of the single photon responses and the fact that even single G proteins produce a detectable response. This made possible the observation that nonhydrolyzable GTP analogues produce discrete transient events rather than the step-like events that would be predicted by previous models. This observation led us to a new view of how enzyme deactivation is coupled to GTP hydrolysis on G protein. According to this view, enzymes are activated by G protein, but can be deactivated by processes that are not dependent on G protein or the hydrolysis of GTP. We have conducted several types of experiments, including some on the vertebrate rod system, that strongly support this hypothesis. A second major theme of this review is transduction noise. The available biochemical evidence suggests that both G protein and G protein-activated enzymes are likely to become spontaneously active and generate undesirable noise. Our measurements indicate, however, that this noise is orders of magnitude smaller than would be predicted by simple models, suggesting that special mechanisms must exist for suppressing this noise. We have proposed a specific mechanism by which enzymes regulated allosterically by multiple subunits could act as coincidence detectors to reduce transduction noise. Finally, there is the fundamental question of which second messengers have a direct role in invertebrate phototransduction. After Fesenko et al. (1985) showed that the light-dependent conductance in vertebrate rods was modulated by cGMP and not by Ca2+, there was rapid progress in understanding the vertebrate photoreceptor transduction mechanism. Now that it has been established that invertebrate light-dependent channels are regulated by cGMP and not by Ca2+, we can expect rapid progress in understanding invertebrate phototransduction. A key question that needs to be answered is whether the InsP3-Ca2+ pathway somehow triggers changes in cGMP or whether there is an altogether different pathway by which cGMP metabolizing enzymes are affected by light.