This article addresses the response of failed ceramics. Under high-velocity impact, ceramics transition from a solid intact material to a fragmented and granular material. This process is often referred to as “damage and failure” and is a complex phenomenon. Because ceramics are very strong in compression, it is difficult to perform laboratory experiments that produce conditions similar to those produced during projectile impact, where the ceramic transitions from an intact material to a granular (failed) material. This limitation generally requires the damage and failed strength to be inferred from computed results that provide good agreement with ballistic penetration experiments. Previous work by the authors [J. Appl. Phys. 97, 093502 (2005)] has suggested a relatively low failed strength for silicon carbide (∼200MPa) that is generally lower than other published data (although the data vary significantly). Work presented here provides additional evidence for a low failed strength for silicon carbide (and also aluminum nitride and boron carbide). Experimental and computed results of high-velocity penetration into thick ceramic targets exhibit large after-flow penetration (the difference between primary penetration and total penetration) that is strongly influenced by the strength of the material directly in front of the penetrator. The large after-flow observed in the experiments and computed results are consistent with a low failed strength. Similar behavior is also observed for aluminum nitride and boron carbide, suggesting that the failed strength of ceramics may be less a function of the specific material and more a characteristic of granular flow under the conditions of high-velocity impact. To provide additional insight into the response of granular material, an analysis of recent ballistic experiments into silicon carbide powder was performed, where the strength of the powder was determined - from the computed results. The analysis indicated that the powder has significant strength under high-velocity impact, much higher than the 200 MPa determined for the failed strength. This suggests that there is something significantly different between the response of powder and that of failed material in the Johnson–Holmquist–Beissel (JHB) model. Possible explanations include, inaccuracies in the computational and numerical techniques, inaccuracies with the JHB model, the tested failed material is not the same as that which exists in the ballistic tests, and/or the failed material exhibits phenomena not included in the JHB model.