To assess the feasibility of obtaining reliable multifocal rod electroretinograms (ERGS) and to compare them to full-field ERGs.Multifocal rod ERGs were recorded using a stimulus array of 61 hexagons. The minimum number of dark, blank frames between flashes was varied from 0 (a minimum of 13.3 msec between flashes) to 21 (a minimum of 293 msec between flashes). Full-field ERGs were obtained using trains of flashes designed to simulate the multifocal sequences. Flashes were blue (W47B), except in a few cases in which red (W26) was used to check for cone intrusion. Flash intensities varied from -1 to 1.7 log scot td-s.Dark-adapted, multifocal ERGs to blue flashes had a small, early component followed by a larger, late component. The early component showed little change in amplitude with increasing intensity. Comparisons with the full-field ERGs indicated that the early component was the focal response. The larger, late component was the response to stray light, and it can be suppressed with the addition of a surround. The focal response was from a relatively circumscribed retinal region. This is shown by comparing the multifocal rod responses from a patient with retinitis pigmentosa to her behaviorally measured rod visual field.By choosing conditions (namely, flashes of moderate intensity with a surround) to minimize the effects of stray light, multifocal rod ERGs can be recorded with sufficient localization to be clinically useful. However, the signal-to-noise ratio of these multifocal rod ERGs was poorer than for multifocal cone responses for comparable recording periods because of the need for blank frames and the slower recovery of the rods to successive presentations.