For efficient production of hydrocarbons from subsurface reservoirs it is important to understand the spatial properties of the reservoir. As there is almost always too little information on the reservoir to build a representative model directly, other techniques have been developed for generating reservoir models. Different assumptions about the distribution of, and relationships between these properties are required for each different technique. However, it is poorly understood how these assumptions, and errors thereof, affect the predictive capability of the resulting reservoir models. In this study these relationships are analyzed from several different perspectives.