The pattern ERG (PERG) is used as an indicator of retinal ganglion cell function. Up to now, reports on the reproducibility of the PERG have been contradictory. We investigated the reproducibility under the conditions of the forthcoming ISCEV guidelines for the PERG. We simultaneously recorded the PERG and VEP in 42 eyes of 21 subjects to phase-reversing checkerboard stimuli with DTL electrodes. Both transient (2 rps) and steady-state (16 rps) stimulation was employed. The check sizes were 0.4 degree, 0.8 degree and 16 degrees, the mean luminance 45 cd/m2, the contrast 98%, and the field size 32 degrees x 27 degrees. Measurements were repeated at the same time of day after 1 week. In addition, we compared two different electrode positions in 16 eyes: (1) across the cornea along the lower lid; and deep in the conjunctival sac. With position (1) the amplitudes were found to be higher by 20% than (2). We calculated the coefficient of variation (CV) of amplitude as a measure of reproducibility. CV was 7 +/- 1% for the steady-state PERG, 9 +/- 1% for the transient PERG, 12 +/- 2% for the steady-state VEP and 14 +/- 3% for the transient VEP. For the latency of the PERG, the intersession CV was found to be 1.5%. Amplitude reproducibility was somewhat higher under steady-state as compared to transient stimulation; we attribute this to the high noise rejection of the Fourier analysis. Altogether, the amplitude reproducibility of the PERG is somewhat higher than that of the VEP.