Is there predictive value in water computer simulations?

There are hardly more important and less understood liquids than liquid water and aqueous solutions. The description of the thermodynamics and kinetics of uniform and, especially, non-uniform aqueous systems is the central problem of contemporary liquid state chemical physics with important applications in biochemistry, electrochemistry, industrial chemistry and geophysics. The importance of water and aqueous solutions explains the interest in their description by computer simulation methods such as Monte Carlo and molecular dynamics. We discuss the actual question of the limits of applicability of currently popular computational models. It is stressed that hydrogen bonding, one of the most important interactions in nature, cannot be quantitatively described in the framework of empirical potential models.