Abstract
The collision frequency in liquids is computed with a cell model and with an exact formula. The results suggest that the cell model is qualitatively accurate in the region of high density if a particular definition of collision is used; it is not appropriate for estimating the frequency of high‐energy collisions.