Following unsuccessful attempts to record visual-evoked potentials (VEPs) in dogs with scalp electrodes, adoption of a new stimulation technique seems to be beneficial. Previously, flashes of white light administered after dark adaptation induced relatively high amplitude electroretinograms (ERGs) covering any VEP activity over the surface of the skull. ERG amplitude, however, can be significantly reduced using flashes of red light after light adaptation (mostly cone stimulation). Simultaneous ERG and VEP recording allows identification of VEPs composed of three significantly different negative peaks (N1, N2, and N3) measured in dogs anesthetized with chloralose and halothane. No more than two of the three peaks were seen in one recording. Only the N1 and N3 waves were consistently recorded in dogs anesthetized with thiopental and thiopental combined with halothane. In 50% of all recordings, N1 was seen alone. The other VEPs consisted of N1 and N2, or N1 and N3 occurring concurrently. The simultaneous occurrence of N2 and N3 waves, however, was never seen. Among all recordings, N1 was most frequently recorded (85% of measurements), followed by N3 and N2 (38% and 31% of measurements, respectively). Peaks of less than 90 ms are highly reproducible. Anesthesia is necessary to eliminate frequent artifacts obtained in conscious and sedated dogs. Thiopental and/or halothane had no effect on measured latencies compared with chloralose.