The infrared spectra of ice and liquid water and the interpretation of them is reviewed. The best understood band is the translational lattice band in ice below 350 cm-1, largely because the translational lattice vibrations can be taken to be mechanically regular in a first approximation, but electrically irregular owing to the orientational disorder. This allows a relatively simple dynamical and electrical theory to be worked out that is necessary for the interpretation. The rotational vibrational and the O—H stretching bands are less well understood, and there is essentially no detailed interpretation that can be reasonably maintained. The spectrum of the liquid is even less well understood, except in the most general terms, and, contrary to the conclusions of a number of papers, there is little reasonably well based structural information to be obtained at present.