Interpreted geologic domains provide the geologic architecture of many current models used to estimate the potential economic resources in mineral deposits. The spatial extent of each domain is usually defined by a three-dimensional wireframe or triangulated surface.The set of geologic wireframes forming part of a resource model might generally be called a spatial interpretation of some set of geologic features coincident with the mineralisation, based primarily on logged and mapped properties of the samples and other exposures of the mineralisation. The interpreted spatial distribution of rock type, alteration and grade of drill hole samples above a particular grade threshold are commonly used criteria in constructing the geologic domain wireframes. Frequently, the grade threshold is closely related to the anticipated economic cut-off grade. The boundaries or spatial limits of different domains are most commonly defined as hard' or soft'. A hard' boundary usually means no dependent relationship of any kind is defined between the properties of samples inside a particular domain wireframe and those samples outside the wireframe. A soft' boundary commonly implies that some kind of dependent relationship does prevail between the properties of samples inside and outside the particular wireframe. Many ways are used to describe those relationships, few of which are based on spatial analysis of the continuity of sample attributes across the domain boundaries. In local estimation procedures such as ordinary block kriging, hard' boundaries act as a total barrier to the influence of samples on either side of the boundary depending on the location of the block being estimated. In some instances, semi-permeable' boundaries are used in which the samples inside a domain wireframe are allowed to influence the estimates outside but not the reverse. This can lead to many undesirable effects in estimates of block grades.This paper discusses the value of this approach to geologic domaining from a viewpoint of model complexity, as well as in relation to the implications for both linear and non-linear estimation methods and contrasts it with alternatives. It also explores some important consequences of using hard' and soft' geological boundaries with the commonly used modelling methods: excessive smoothing of estimates near domain boundaries and conditionally biased estimates of grade at the global and local scale.