1. It is widely believed that signals contributing to the receptive field surrounds of retinal bipolar cells pass from horizontal cells to bipolar cells via GABAergic synapses. To test this notion, we applied gamma-aminobutyric acid (GABA) agonists and antagonists to isolated, perfused retinas of the salamander Ambystoma tigrinum while recording intracellularly from bipolar cells, horizontal cells, and photoreceptors. 2. As we previously reported, administration of the GABA analogue D-aminovaleric acid in concert with picrotoxin did not block horizontal cell responses or the center responses of bipolar cells but blocked the surround responses of both on-center and off-center bipolar cells. 3. Surround responses were not blocked by the GABA, antagonists picrotoxin or bicuculline, the GABAB agonist baclofen or the GABAB antagonist phaclofen, and the GABAC antagonists picrotoxin or cis-4-aminocrotonic acid. Combinations of these drugs were similarly ineffective. 4. GABA itself activated a powerful GABA uptake mechanism in horizontal cells for which nipecotic acid is a competitive agonist. It also activated, both in horizontal cells and bipolar cells, large GABAA conductances that shunted light responses but that could be blocked by picrotoxin or bicuculline. 5. GABA, administered together with picrotoxin to block the shunting effect of GABAA activation, did not eliminate bipolar cell surround responses at concentrations sufficient to saturate the known types of GABA receptors. 6. Surround responses were not blocked by glycine or its antagonist strychnine, or by combinations of drugs designed to eliminate GABAergic and glycinergic pathways simultaneously. 7. Although we cannot fully discount the involvement of a novel GABAergic synapse, the simplest explanation of our findings is that the primary pathway mediating the bipolar cell's surround is neither GABAergic nor glycinergic.