A 15-sec stimulus followed by unavoidable monetary loss was presented to human subjects who were avoiding loss on a free-operant schedule. As has been observed in studies where shock is the aversive event, initial reactions to the pre-loss stimulus were transient increases in overall and stimulus rates. Unlike shock studies, continued training produced decreased rates, in the presence of the 15-sec stimulus, which were maintained in two of three subjects. Subsequent observations indicated that lowered rates were a function of the subject's rate of avoidance responding, the duration of the stimulus, and the scheduling of avoidable losses. Increasing the duration of the stimulus eliminated lowered rates in the presence of the stimulus and subsequent exposures to conditions which previously produced lowered rates did not result in recovery of the phenomenon. Introduction of the pre-loss stimulus on an extinction baseline (avoidable losses were omitted), however, reinstituted lowered rates. It is proposed that the pre-loss stimulus assumed discriminative control over low rates because responding in the presence of the stimulus was ineffective in avoiding the unavoidable loss. Recovery from lowered rates is attributed to the occurrence of avoidable losses during the stimulus period, and maintenance of lowered rates on the extinction schedule to the omission of such avoidable losses.Free-operant avoidance behavior of human subjects can be effectively maintained by "loss" of or "timeout" from positive reinforcement. Stone (1961), for example, trained subjects to avoid the disappearance of pennies from a magazine displayed before them. If the subject did not respond, pennies disappeared at a fixed rate; each response postponed the next disappearance of a penny for a fixed time period. In a similar manner, Baron and Kaufman (1966) studied the aversive properties of timeout from monetary reinforcement using a schedule in which each response postponed termination of a payment signal. Other experiments have demonstrated the aversive properties of such events as loss of points (Weiner, 1963) and timeout from a cartoon movie for nursery school children (Baer, 1960 rather than electric shock, as the aversive event. For this reason, they provide relatively little systematic information about the variables controlling human avoidance of loss or timeout. A notable exception, however, is the finding that avoidance rates increase as an inverse function of the interval by which loss or timeout is postponed (Baron and Kaufman, 1966;Stone, 1961).The present study was designed to identify other variables controlling loss-avoidance by human subjects. The procedures were suggested by studies in which animal subjects were trained to avoid shock, in particular, a study by Sidman, Herrnstein, and Conrad (1957). In that experiment, while monkeys were avoiding electric shock, a stimulus was concurrently presented and followed by unavoidable shock. Heightened rates of avoidance behavior resulted with the most marked increases occurring in the ...