The shapes of the photocurrent responses of rat rods, recorded with microelectrodes from the receptor layer of small pieces of isolated retinas, have been investigated as a function of temperature and of stimulus energy. Between 27 and 37 degrees C the responses to short flashes can be described formally as the output of a chain of at least four linear low-pass filters with time constants in the range 50-100 msec. The output of the filter chain is then distorted by a nonlinear amplitude-limiting process with a hyperbolic saturation characteristic. Flashes producing approximately 30 photons absorbed per rod yield responses of half-maximal size independently of temperature. The maximum response amplitude is that just sufficient to cancel the dark current. The rate of rise of a response is proportional to flash energy up to the level of 10(5) photons absorbed per rod, where hyperbolic rate saturation ensues. The responses continue to increase in duration with even more intense flashes until, at the level of 10(7) photons absorbed per rod, they last longer than 50 min. The time-courses of the photocurrent and of the excitatory disturbance in the rod system are very similar. The stimulus intensity at which amplitude saturation of the photocurrent responses begins is near that where psychophysical "rod saturation" is seen. An analysis of these properties leads to the following conclusions about the mechanism of rod excitation. (a) The kinetics of the photocurrent bear no simple relation to the formation or decay of any of the spectroscopic intermediates so far detected during the photolysis of rhodopsin. (b) The forms of both the amplitude- and rate-limiting processes are not compatible with organization of rhodopsin into "photoreceptive units" containing more than 300 chromophores. Even at high stimulus intensities most rhodopsin chromophores remain connected to the excitatory apparatus of rods. (c) The maximum rate of rise of the photocurrent is too fast to be consistent with the infolded disks of a rod outer segment being attached to the overlying plasma membrane. Most of the disks behave electrically as if isolated within the cell. (d) Control of the photocurrent at the outer segment membrane is not achieved by segregation of the charge carriers of the current within the rod disks. Instead, it is likely to depend on control of the plasma membrane permeability by an agent released from the disks.