Successful modeling of flow and transport in geologic formations requires detailed characterization of their hydraulic properties. In practice, these properties are at best measured at selected locations where their values are often corrupted by experimental and interpretive errors. Such data are further used in model parameterization and subsequent numerical simulations. We investigate the propagation of measurement errors through all stages of our modeling process: (1) data collection mimicked by numerical simulations of transient pumping tests; (2) transient semilog data analysis; (3) geostatistical generation of permeability distributions with nugget; and, finally, (4) flow simulations. We demonstrate how even small experimental and interpretive errors can manifest themselves in inaccurate predictions of the system states.