It is well known that metals and alloys erode at lower impact threshold velocities than expected. This was studied by Thomas and Brunton (1970) who reported “discrepancy factors” in the range 4–10. These authors suggested that liquid impact was a more searching impact than solid impact since it was able to exploit weaknesses in the metal. Further, as erosion develops the sideways liquid flow can add shear stresses to surface steps and hydraulically load liquid trapped in cracks and crevices. In 1970, Brunton and Camus showed that during the impact process cavities could form inside the liquid drops, and that some collapsed onto the solid surface. This provided a second potential mechanism to explain the low damage threshold with the cavity collapse adding to the “water hammer” pressures. However, Brunton and Camus were cautious in offering this as a potential mechanism. In this paper, we argue that the Brunton and Camus experiments were in a relatively low velocity regime (20–70 m s−1), compared with those in turbine erosion and rain erosion of aircraft components. This paper presents high-speed photographic sequences of cavity formation and shock propagation in impacted liquids using a range of techniques. Finally, a new method is described in which a metal projectile is fired at a liquid jet. This produces large amounts of cavitation. Our conclusion is that the cavity collapse process can explain the observed lower threshold velocities