Individuals with anxiety disorders often do not respond to safety signals and hence continue to be afraid and anxious. Consequently, it is important to develop paradigms in animals that can directly study brain systems involved in learning about, and responding to, safety signals. We previously developed a discrimination procedure in rats of the form AX+/BX-, where cues A and X presented together are paired with an aversive stimulus and cues B and X presented together predict the absence of an aversive stimulus. The present experiment adapted this procedure to the fear-potentiated startle paradigm in rhesus monkeys.