Abstract The erosion of glass as a function of the velocity of alumina particles of 30μm average size was studied. It was found that the use of the rotating double-disk method to determine particle velocity can lead to systematic deviations. Correcting for this effect, the results indicate that the erosion can be well described using indentation fracture mechanics. The exponent of the velocity-dependence of the removal rate was found to be in good agreement with the value of 7 3 which is predicted by quasi-static indentation theory. Two transitions have been observed: at low velocities the removal process changes from ductile to brittle, while at high velocities fragmentation of the alumina particles sets in. The corresponding impact loads can also be obtained from indentation theory and are in line with the observed effects.