1. In the Xenopus retina, the effects of selective D1 and D2 dopamine ligands on photoreceptor to horizontal cell transfer were studied by intracellular recording from horizontal cell axons. Rod and cone inputs to the horizontal cell were estimated by adjusting the intensities of red and green flashes to elicit equal rod tails. The resultant waveforms were digitized and subtracted, and their difference was taken to reflect solely cone input to the horizontal cell. 2. It was found that both D1 (SKF 38393) and D2 (LY 171555) agonists increased the amplitude and quickened the kinetics of cone-to-horizontal cell transfer; they also depolarized the horizontal cell by 8-10 mV. In contrast, either D1 or D2 agonists reduced the rod input to the horizontal cell without altering its kinetics. 3. Type D2 antagonists reduced and slowed the cone input and hyperpolarized the horizontal cell. D2 antagonists increased the rod input but left its kinetics unchanged. 4. Although both D1 and D2 agonists elicited qualitatively similar effects, the D1 agonist evoked a greater increase in the amplitude and a greater acceleration of the kinetics of the cone input than did the D2 agonist. Moreover, the action of the D1 agonist was blocked by SCH 23390 but not by spiroperidol or metoclopramide, whereas the reverse was true for the D2 agonist. These data indicate that D1 and D2 agonists probably act at different sites. 5. The pharmacologic findings are interpreted to indicate that dopamine ligands act primarily through the cone pathway and that rod-to-horizontal cell transfer is shunted to a variable degree. 6. An equivalent circuit model was developed for a spine-bearing portion of a horizontal cell axon of the Xenopus retina. Anatomic study shows that such spines branch, making contact with both rod and cone photoreceptor bases. Thus there are two conductance pathways in parallel for rod-to-horizontal cell and cone-to-horizontal cell transmission. The model is used to test the hypothesis that mutual shunting in the two pathways can account for the physiological effects observed. 7. The values of the purely resistive elements of the pathway are based on their dimensions. Membrane resistance was taken to be 5,000 omega/cm2 and axial resistance 200 omega/cm. The photoreceptor-to-horizontal cell synaptic battery was taken to be composed of glutamate-sensitive channels, with unitary channel conductance of 6 pS. Channel density was estimated from freeze-fracture data at 5,000 microns-2. A potassium battery and a glycine-sensitive synaptic input from an interplexiform cell were modeled to exist in parallel with the light-sensitive battery. 8. Dopamine was assumed to increase the conductance of the cone-to-horizontal cell synapse, but not to affect the conductance of the rod-to-to-horizontal cell synapse, consistent with physiological measures.(ABSTRACT TRUNCATED AT 400 WORDS)