Groundwater flow and transport take place in complex geologic environments whose makeup cannot be ascertained in all of its relevant details. This makeup tends to exhibit discrete and continuous variations on a variety of scales, causing hydraulic parameters to do likewise. In practice, such parameters are at best measured at selected locations where their values depend on the scale and method of measurement. The data, affected by experimental and interpretive errors, are further used in model parameterization. This paper continues on our previous development of mathematical quantification of measurement errors and error propagation. We demonstrate that geostatistical techniques used to construct permeability fields do not influence the propagation of errors. At the same time, the difference between the flow prediction based on the data corrupted by measurement errors and the flow prediction based on "true" permeability values can be significant.