The whelk Nucella lamellosa displays phenotypic plasticity in the presence of the sea star Pisaster ochraceus by becoming more retractable. In this study, we directly tested whether this response is an inducible defense, and looked for evidence of phenotypic costs associated with the induced phenotype. We found that whelks held in the presence of sea stars consuming conspecific whelks became more retractable, while whelks that were not exposed to sea stars became less retractable—indicating that this is a reversible and symmetric (i.e. similar magnitude of change) response. We did not find changes in aspect ratio of the shell or size of the whelks. Following the induction experiment, whelks were fed to sea stars. In this predation experiment, sea stars were much less likely to consume whelks previously exposed to sea stars compared to whelks not exposed to sea stars. There was a strong relationship between mortality and retractability relative to shell length, and individuals that could retract 50% of their shell length had relatively little chance of being consumed during the predation experiment. These results support the hypothesis that increased retractability is an inducible defense. We also conducted field surveys of N. lamellosa populations and found differences in retractability among populations, most of which were more retractable than the whelks in our induction experiment. For these field-collected individuals, we found evidence of phenotypic costs, with a negative relationship between retractability and tenacity. Thus, N. lamellosa responds to a sea star predator by becoming more retractable but at the cost of becoming less tenacious.