Distributed models of solute transport at the field and catchment scales require the specification of effective parameters at the model grid scale. The grid scale will generally be much larger than the scale at which it is possible to make measurements to derive parameter values but may be of the same order as the scale of variability of such ‘point’ values. In addition, measurements are usually expensive and time consuming to make and where, for example, ‘undisturbed’ soil cores are used, may be destructive. This paper describes a conditional probability based approach for estimating grid scale effective parameter values in the light of expected spatial heterogeneity, given only one or a small number of available measurements. Initial results show, somewhat surprisingly, that despite the integrative nature of the spatial averaging involved in moving to the grid scale, the variance of the effective grid element values does not decrease. There is a small shift in the location of the distribution, as a result of the macrodispersive effect of the grid scale variability.