This report deals with Kriging, a spatial interpolation-method that enables making predictions of the value of a measurable property on locations where this property has not actually been measured. The imprecision of these predictions can also be assessed. The relation is described between Kriging (also referred to as Optimum Interpolation) and some conventional statistical techniques. Non-stationary spatial patterns can be modelled using Intrinsic Random Functions. References are given to extensions of Kriging with time as an additional dimension. The theoretical basis of specification of uncertainty in the predicted values following from uncertainty in the variogram, is as yet unsatisfactory.<br>