By both tradition and purpose, the land parameterization schemes of hydrological and meteorological models differ greatly. Meteorologists are concerned primarily with solving the energy balance, whereas hydrologists are most interested in the water balance. Meteorological climate models typically have multi-layered soil parameterisation that solves temperature fluxes numerically with diffusive equations. The same approach is carried over to a similar treatment of water transport. Hydrological models are not usually so interested in soil temperatures, but must provide a reasonable representation of soil moisture to get runoff right. To treat the heterogeneity of the soil, many hydrological models use only one laver with a statistical representation of soil variability. Such a hydrological model can be used on large scales while taking subgrid variability into account. Hydrological models also include lateral transport of water - an imperative if river discharge is to be estimated. The concept of a complexity chain for coupled modelling systems is introduced, together with considerations for mixing model components. Under BALTEX (Baltic Sea Experiment) and SWECLIM (Swedish Regional Climate Modelling Programme), a large-scale hydrological model of runoff in the Baltic Basin is used to review atmospheric climate model simulations. This incorporates both the runoff record and hydrological modelling experience into atmospheric model development. Results from two models are shown. A conclusion is that the key to improved models may be less complexity. Perhaps the meteorological models should keep their multi-layered approach for modelling soil temperature, bur add a simpler, yet physically consistent, hydrological approach for modelling snow processes and water transport in the soil.