1. Physiological noise in the visual transduction mechanism was studied by recording membrane current from single rod outer segments in pieces of isolated toad retina. 2. The inward current in darkness showed spontaneous fluctuations which disappeared during the response to bright light. 3. The dark noise consisted of two components, a continuous fluctuation of rms amplitude about 0.2 pA and occasional discrete events about 1 pA in size. 4. Intervals between discrete events followed the exponential distribution expected of a Poisson process with a mean rate of about one event per 50 sec (20 degrees C). 5. The amplitude and power spectrum of the discrete events resembled those of single photon effects in the same rod, suggesting that discrete events may arise from spontaneous activation of single rhodopsin molecules. 6. The temperature dependence of the mean frequency of occurrence of discrete events gave an activation energy of 22 kcal mole-1, probably characteristic of thermal isomerization of rhodopsin. 7. The variance of the continuous component of the dark noise rose linearly with the length of the outer segment drawn into the suction electrode, indicating that this component is generated in the outer segment. 8. The power spectrum of a rod's continuous noise was usually fitted by the square of a Lorentzian with the same time constant as that of the four first-order delays in the cell's single photon response. The shot effects composing the continuous component thus appear to be shaped by two of four sequential processes in transduction. 9. The variance and spectrum of the continuous noise are interpreted to reflect shot effects about 1/400 the size of a single photon effect occurring at a frequency of 6 x 10(3) sec-1. 10. The rod's flash sensitivity was halved by a steady light to giving about 8 photoisomerizations sec-1. The much lower mean rate of discrete events indicates that Io in increment sensitivity experiments on individual receptors is not set by thermal activation of rhodopsin. 11. Values of sensitivity and time-to-peak flash response collected from many cells in darkness were correlated by the same power law relation obtaining in the presence of backgrounds. The correlation observed would be explained if a single variable controlled both the gain and time scale of several stages of the transduction mechanism in background light and in darkness.