Safety signals provide "relief" through predicting the absence of an aversive event. At issue is whether these signals also act as instrumental reinforcers. Four experiments were conducted using a free-operant lever-press avoidance paradigm in which each press avoided shock and was followed by the presentation of a 5-sec auditory safety signal. When given a choice between two levers in Experiment 1, both avoiding shock, rats preferentially responded on the lever that produced the safety signal as feedback, even when footshock was omitted. Following avoidance training with a single lever in Experiment 2, removal of the signal led to a decrease in avoidance responses and an increase in responses during the safety period normally denoted by the signal. These behavioral changes demonstrate the dual conditioned reinforcing and fear inhibiting properties of the safety signal. The associative processes that support the reinforcing properties of a safety signal were tested using a novel revaluation procedure. Prior experience of systemic morphine during safety signal presentations resulted in an increased rate of avoidance responses to produce the safety signal during a drug-free extinction test, a finding not seen with d-amphetamine in Experiment 3. Morphine revaluation of the safety signal was repeated in Experiment 4 followed by a drug-free extinction test in which responses did not produce the signal for the first 10 min of the session. Instrumental avoidance in the absence of the signal was shown to be insensitive to prior signal revaluation, suggesting that the signal reinforces free-operant avoidance behavior through a habit-like mechanism.