Four groups of rats were conditioned to avoid shock in a two-way shuttlebox. During acquisition. each response both precluded a brief programmed shock and produced a I-sec feedback signal. CS termination was not permitted during either acquisition or' extinction. During extinction, response-produced or yoked feedback was combined with either 100% response-independent shock or no-shock conditions. Greatest resistance to extinction of avoidance behavior was obtained when response-produced feedback occurred in the absence of shock relative to any other combination of these events. These results, when compared to those of previous studies investigating the role of warning signal termination in combination with the presence or absence of shock in extinction (Hartley, 1968), clearly reveal a functional similarity between feedback and CS termination. The most prevalent interpretation of discriminative avoidance behavior has been the two-factor theory, originally discussed by Mowrer and Lamoreaux (1946) and more recently by Solomon and Brush (1956). According to the theory, conditioning is accomplished by both Pavlovian and operant processes. A classically conditioned fear response develops by repeated pairings of CS and US. This fear reaction motivates the subject to avoid the US. Reinforcement of avoidance occurs when fear is reduced by response-produced termination of the CS and preclusion of the US. An alternative explanation of avoidance views the CS, CS termination, and the presence or absence of the US as having discriminative rather than motivational prop erties (Mowrer & Jones, 1945; Sheffield, 1949). When one or all of these cues are altered during extinction, avoidance behavior declines due to generalization decrement. Thus, if a subject is avoiding either a high proportion of shocks during terminal acquisition or all of them, a no-shock extinction procedure should pro long extinction over a procedure in which shocks occur on all extinction trials . Similarly , any extinction pro cedure that alters the consequences of responding from those occurring during the acquisition phase, whether the events be CS termination, US avoidance, or any other highly discriminable events, should facilitate extinction of the behavior.