1. We recorded the a-wave of the human electroretinogram from subjects with normal vision, using a corneal electrode and ganzfeld (full-field) light stimulation. From analysis of the rising phase of rod-isolated flash responses we determined the maximum size (amax) of the a-wave, a measure of the massed circulating current of the rods, and the amplification constant (A) of transduction within the rod photoreceptors. 2. During light adaptation by steady backgrounds the maximal response was reduced, as reported previously. amax declined approximately as I0/(I0 + IB), where IB is retinal illuminance and I0 is a constant. In different subjects I0 ranged from 40 to 100 trolands, with a mean of 70 trolands, corresponding to about 600 photoisomerizations s-1 per rod. (1 troland is the retinal illuminance that results when a surface luminance of 1 cd m-2 is viewed through a pupil area of 1 mm2.) The amplification constant A decreased only slightly in the presence of steady backgrounds. 3. Following a full bleach amax recovered along an S-shaped curve over a period of 30 min. There was no detectable response for the first 5 min, and half-maximal recovery took 13-17 min. 4. The apparent amplification constant decreased at early times after large bleaches. However, upon correction for reduced light absorption due to loss of pigment, with regeneration of rhodopsin occurring with a time constant of 9-15 min in different subjects, it appeared that the true value of A was probably unchanged by bleaching. 5. The recovery of amax following a bleach could be converted into recovery of equivalent background intensity, using a 'Crawford transformation' derived from the light adaptation results. Following bleaches ranging from 10 to > 99 %, the equivalent background intensity decayed approximately exponentially, with a time constant of about 3 min. 6. The time taken for amax to recover to a fixed proportion of its original level increased approximately linearly (rather than logarithmically) with fractional bleach, with a slope of about 12 min per 100 % bleach. Similar behaviour has previously been seen in psychophysical dark adaptation experiments, for the dependence of the 'second component' of recovery on the level of bleaching.