We apply first-principles molecular dynamics and static calculations in the study of several layers of water adsorbed at an oxide surface. Our aim is to bridge the gap between the well-explored monolayer and the complex, little-understood oxide-water interface by probing this middle ground where there is still contact with vacuum surface experiments. By examining coverages up to three monolayers on rutile TiO2(110) we emerge with three major conclusions. First, there is a trend with increasing coverage for water near the surface to become more molecular in nature, viz. the coordination of H to O for all near-surface molecules tends to lower values as the coverage increases. This means the hydroniumlike and other partially dissociated structures predicted up to 1.5 monolayers become less favorable as the coverage increases. Second, the templating influence of the surface disrupts the structure of the second and third layers at 3 ML coverage. Third, we can make a very satisfactory interpretation of experimental results (HREELS and TPD) gaining some new insights in the process. The assignment of TPD peaks to layer-by-layer desorption is shown to be open to question, as some second-layer molecules may contribute to the same peak as do those in the third layer. (C) 2003 American Institute of Physics.