Separate groups of gerbils received shuttle-avoidance training or classical trials (conditioned and unconditioned stimulus [CS and US, respectively] pairings) in a 100-trial session. The shock US was either escapable or inescapable, i.e., of fixed duration (.1-3.0 sec.). Conditioned responses, intertrial interval responses, and observations of the subjects' reactions to each US were recorded. Avoidance learning emerged only in groups exposed to escapable shock or a brief inescapable shock. Based on both the observational data of the nature of the shock-elicited reactions and shuttle performance, it was concluded that response termination of the US is not necessary for shuttle-avoidance learning. The results were discussed in terms of a punishment theory of avoidance and the species-specific defense-reaction hypothesis.