Purpose. Based on recent pharmacological results (Grzywacz, Tootle & Amthor, Vis. Neurosci., In Press), we postulated a two-asymmetric pathway hypothesis for retinal directional selectivity. One of these pathways is mediated by the starburst amacrine cell (SAC) and receives GABAergic and glutaminergic inputs (primarily through AMPA receptors- Cohen & Miller, 1995, PNAS, 92:1127). The output of this pathway is a cholinergic synapse onto the On-Off directionally selective ganglion cell (DS GC). The other pathway comprises asymmetric GABAergic and symmetric glutaminergic (of mixed receptor type- Kittila & Massey, 1995, ARVO) synapses to DS GC. Glutaminergic synapses feed the cell that makes the asymmetric GABAergic synapse primarily through AMPA receptors. Our goal was to test this two-asymmetric-pathway hypothesis. Methods. We investigated the contrast dependence of curare's (cholinergic antagonist) and picrotoxin's (GABAergic antagonist) effects on directionality when delivered in isolation or combination. Results. Although these drugs in isolation could not block directionality completely regardless of contrast, their combination eliminated directional selectivity. Picrotoxin's effects in isolation already suggested more than one asymmetric pathway and that different cells receive different amounts of each pathway. Namely, while for 33% of the cells, picrotoxin eliminated directionality regardless of contrast, for the other 67% some directionality remained even under saturating concentrations of the drug. Curare's effects in isolation also suggested a multiple-asymmetric pathway for directionality. On one hand, curare impaired but failed to eliminate directionality completely. On the other hand, simultaneous recordings from DS GC and cholinergic SAC strongly suggested that the latter makes asymmetric synapses onto the former. Moreover, dendritic fasciculations between DS GC and SAC were spatially asymmetric. Hence, cholinergic synapses appeared to be asymmetric, but were not necessary for directionality. Conclusions. Our results indicate that at least two asymmetric pathways mediate rabbit retinal directional selectivity.