A series of three experiments examined the forgetting of visual discriminations by pigeons. The problems consisted of feature discriminations, with dot displays as the discriminative stimuli, and involved a successive go-no-go pecking response. It was found, in all three experiments, that pigeons trained to refrain from pecking an S- display resumed pecking at this display after retention intervals. It was argued that these data represent the first direct demonstration of forgetting of a discrimination by pigeons. In addition to the simple demonstration of forgetting, it was found in Experiment 1 that the amount of forgetting progressively increased, in a negatively accelerated fashion, over intervals of 1, 10, and 20 days. Also, more S- responses occurred during relearning a reverse discrimination than after relearning a nonreverse discrimination. In Experiment 2, acquisition was retarded and more forgetting occurred for discriminations that involved more highly similar stimuli. In Experiment 3, a change in contextual cues between acquisition and retention testing enhanced forgetting when the contextual cues present during original acquisition were conspicuous; when these cues were relatively inconspicuous, a change in context had no effect on forgetting.