We used suction-pipette recording and fluo-4 fluorescence to study light-induced Ca2+ release from the visible double cones of zebrafish. In Ringer, light produces a slow decrease in fluorescence which can be fitted by the sum of two decaying exponentials with time constants of 0.5 and 3.8 s. In 0Ca2+–0Na+ solution, for which fluxes of Ca2+ across the outer segment plasma membrane are greatly reduced, light produces a slow increase in fluorescence. Both the decrease and increase are delayed after incorporation of the Ca2+ chelator BAPTA, indicating that both are produced by a change in Ca2+. If the Ca2+ pool is first released by bright light in 0Ca2+–0Na+ solution and the cone returned to Ringer, the time course of Ca2+ decline is much faster than in Ringer without previous light exposure. This indicates that the time constants of 0.5 and 3.8 s actually reflect a sum of Na+/Ca2+-K+ exchange and light-induced release of Ca2+. The Ca2+ released by light appears to come from at least two sites, the first comprising 66% of the total pool and half-released by bleaching 4.8% of the pigment. Release of the remaining Ca2+ from the second site requires the bleaching of nearly all of the pigment. If, after release, the cone is maintained in darkness, a substantial fraction of the Ca2+ returns to the release pool even in the absence of pigment regeneration. The light-induced release of Ca2+ can produce a modulation of the dark current as large as 0.75 pA independently of the normal transduction cascade, though the rise time of the current is considerably slower than the normal light response. These experiments show that Ca2+ can be released within the cone outer segment by light intensities within the physiological range of photopic vision. The role this Ca2+ release plays remains unresolved.