The flash-induced charge movements during the photocycle of light-adapted bacteriorhodopsin in purple membranes attached to a black lipid membrane were investigated under voltage clamp and current clamp conditions. Signal registration ranged from 200 ns to 30 s after flash excitation using a logarithmic clock, allowing the equally weighted measurement of the electrical phenomena over eight decades of time. The active pumping signals were separated from the passive system discharge on the basis of an equivalent circuit analysis. Both measuring methods were shown to yield equivalent results, but the charge translocation could be accurately monitored over the whole time range only under current clamp conditions. To describe the time course of the photovoltage signals a model based on distributed kinetics was found to be more appropriate than discrete first order processes suggesting the existence of conformational substates with distributed activation energies. The time course of the active charge displacement is characterised by a continuous relaxation time spectrum with three broad peaks plus an unresolved fast transient (<0.3 mus) of opposite polarity. The time constants and relative amplitudes (in brackets) derived from the peak rate constants and relative areas of the three bands are: tau(1) = 32 mus (20%), tau(2) = 0.89 ms (15%) and tau(3) = 18 ms (65%) at 25 degrees C in 150 mM KCl at pH7. The Arrhenius plots of the peak rate constants were linear yielding activation energies of E(A1) = 57 kJ/mol, E(A2) = 52 kJ/mol, and E(A3) = 44 kJ/mol. The electrical signal at 890 mus has no counterpart in the photocycle of bacteriorhodopsin suspensions. Fits with a sum of exponentials required 5 to 6 components and were not reproducible. Analysis of photoelectrical signals with continuous relaxation time spectra gave equally good fits with fewer parameters and were well reproducible.