This paper considers the mechanical response of Bombyx mori silk cocoons to knife stabbing, a simple but controlled way of simulating predaceous penetration. Here, we stab test both entire cocoons (EC) and cocoon wall segments (CWS) statically and dynamically, and note that the process can be broken down in three stages. The first stage involves material deflection, the second is knife penetration, and the third is knife perforation. We find that ca. 95 % of the kinetic energy is lost during the penetration stage. There are noticeable differences in strain between the equatorial (strain = 13 %) and meridional (strain = 1.5 %) directions before and after the stabbing of EC specimens (p < 0.001). The apparent area of the cocoon is noted to be on average 7 % lower after stabbing than it is prior to being stabbed (p < 0.01). We find that while compression of the cocoon from stabbing results in equatorial expansion (with a Poisson's ratio, nu = 0.25), in the meridional direction the cocoon contracts (nu = -0.05) thus showing auxetic behaviour. Force-deflection curves are different in CWS specimens as compared to EC specimens, and this is attributable to natural curvatures in CWS specimens remaining even after a being flattened for mounting and testing. Differences between EC and CWS specimens are also noticeable in the sizes of the stab footprints, with EC samples exhibiting 33 % smaller footprints than CWS samples (p < 0.001). We conclude that testing whole cocoon structures provides a more accurate understanding of their properties as compared to cut and flattened structures. This is because flattening cocoon wall specimens induces delamination and multiple failure zones, reducing the natural stab resistance of the material.