Certain unconditioned stimuli (UCS) in flavor avoidance learning sometimes become ineffective after pairings with relatively stronger UCS. This failure of avoidance learning (avfail) has been demonstrated only with rodents. The present investigations were conducted to determine whether avfail might also occur with avian species, the food selection of which is guided primarily by visual cues. In Experiment 1, starlings were given pairings of methiocarb (a relatively weak UCS) and LiCl (a relatively strong UCS). In Experiment 2, red-winged blackbirds were given pairings of two concentrations of methiocarb (relatively weak and relatively strong UCS, respectively). Pairings were followed by a conditioning trial (UCS gavage in the presence of a color cue) and two-choice tests. Conditioned avoidance was always observed except (a) when methiocarb preceded LiCl and (b) when the low preceded the high methiocarb dose in preconditioning pairings. Experiment 3 demonstrated that UCS habituation could not account for the results of Experiments 1 and 2. The data reflect avfail in the visual modality, and a biological implication of the results is that birds may not learn strong avoidance of aposematic prey containing varied levels of toxicant.Pentobarbital normally elicits weak, eas-sickness (Parker, 1979) that summates ily extinguished conditioned flavor avoid-with the mild sickness normally produced ance, whereas lithium chloride usually is a by pentobarbital. The opposite frequently more effective unconditioned stimulus occurs, and pentobarbital loses its capacity (UCS). When pentobarbital precedes LiCl to produce any measurable flavor avoidance administration, one reasonable prediction (Revusky, Taukulis, Parker, & Coombes, (Revusky, Taukulis, & Peddle, 1979) is that 1979; Revusky, Taukulis, & Peddle, 1979). pentobarbital will become a more effective This phenomenon, termed avfail, has been UCS because it elicits conditioned lithium demonstrated with a number of drug-drug pairings (Revusky, Coombes, & Pohl, 1982)