1. Membrane currents initiated by intense, 20 microseconds flashes (photocurrents) were recorded from isolated salamander rods by combined extracellular suction electrodes and intracellular tight-seal electrodes either in current or voltage clamp mode. The magnitudes (mean +/- 2 S.E.M.) of the maximal photoresponses recorded by the suction and by the intracellular electrode respectively were 40 +/- 5 pA (n = 18) and 35 +/- 7 mV (n = 8) for current clamp at zero current; 43 +/- 9 pA and 66 +/- 13 (n = 11) pA for voltage clamp at the zero-current holding potential, -24 +/- 3 mV. 2. Photocurrents initiated by flashes isomerizing 0.1% or more of the outer segment's rhodopsin achieved a saturated velocity and were 95% complete in less than 50 ms. The effect of incrementing flash intensity above 0.1% isomerization can be described as a translation of the photocurrent along the time axis towards the origin. Within the interval 0-50 ms the latter two-thirds of the velocity-saturated photocurrent is well described as a single-exponential decay. The decay was much faster in voltage clamp (2.8 +/- 1.2 ms, n = 11) than in current clamp mode (17 +/- 5 ms, n = 17). 3. The initial third of the velocity-saturated photocurrent, occurring over the interval from the flash to the onset of exponential decay, followed about the same time course in current and voltage clamp. The time interval occupied by this initial 'latent' phase decreased with increasing flash intensity and attained an apparent minimum of about 7 ms in response to flashes isomerizing 10% or more of the rhodopsin at ca. 22 degrees C. 4. The hypothesis that the decay of outer segment light-sensitive membrane current is the same in current and voltage clamp was supported by an analysis of the difference between outer segment currents measured successively in the two recording modes. First, the tail of the difference current decayed exponentially with a time constant approximately equal to R x C, where R and C are independently estimated slope resistance and capacitance of the rod. Secondly, the integral of the difference current, when divided by outer segment capacitance, closely approximated the hyperpolarizing light response measured under current clamp. Thus, displacement current accounted for the difference between photocurrents measured in current and voltage clamp.(ABSTRACT TRUNCATED AT 400 WORDS)