This study measured effects of context reinforcement on a visual discrimination. Five pigeons responded to 1 key in the presence of 6 shorter wavelengths and to another key for 6 longer wavelengths. Psychometric functions provided measures of discriminative sensitivity (d′) and overall stimulus control. Sensitivity and control were slightly but significantly better when 20% of correct choices yielded reinforcement than when 5% did. Reinforcement of pecks to the sample stimulus reduced control substantially and sensitivity slightly; noncontingent reinforcement during intertrial intervals also reduced control, whether such reinforcement was signaled or not. Accuracy was excellent during an extinction session, but it fell substantially when reinforcement for sample pecks was added during choice extinction. Simulations based on a memory model of the discrimination process reproduced most of the experimental findings.