1. The a-wave of the electroretinogram was recorded from human subjects with normal vision, using a corneal electrode and ganzfeld stimulation. We applied the paired-flash technique, in which an intense 'probe' flash was delivered at different times after a 'test' flash. The amplitude of the probe-flash response provided a measure of the circulating current remaining at the appropriate time after the test flash. 2. We extended previous methods by measuring not at a fixed time, but at a range of times after the probe flash, and then calculating the ratio of the 'test-plus-probe' response to the 'probe-alone' response, as a function of time. 3. Under dark-adapted conditions the rod response derived by the paired-flash technique (in response to a relatively dim test flash) peaked at ca 120 ms, with a fractional sensitivity at the peak of ca 0.1 Td(-1) s(-1). 4. As reported previously, background illumination reduced the maximal response, reflecting a reduction in rod circulating current. In addition, it shortened the time to peak (to ca 70 ms at an intensity of 170 Td), and reduced the flash sensitivity measured at the peak. The flash sensitivity declined approximately according to Weber's Law, with a 10-fold reduction occurring at an intensity of 100-200 Td. We could not reliably measure responses at significantly higher background intensities because the circulating current became so small. 5. In order to investigate the phototransduction process after correction for response compression, we expressed the derived response as a fraction of the maximal response that could be elicited in the presence of the background. The earliest rising phase of this 'fractional response per unit intensity' was little affected by background illumination, suggesting that the amplification constant of transduction was unaltered by light adaptation.