A hydraulic field test program was performed at a hard rock laboratory (Aspö HRL) on the Swedish east coast to test upscaling theories. The test program investigated the rock volume around a borehole located at a depth of approximately 340 m below sea level. Hydraulic packer tests were performed at various scales, from 2 m to the entire borehole length of 296 m. From this set of data the predictive ability of different upscaling methods could be evaluated. The comparison of the evaluated "true" field scale hydraulic conductivity with the upscaled hydraulic conductivity yielded that the majority of the upscaling methods tested in this paper predict the large scale values with significant accuracy. However, the ability to predict rapidly decreases when the variance of the natural logarithm of hydraulic conductivity of the subsamples is larger than one. Such a variance is consistently found in the crystalline rocks at the tested site at the 2 m scale. However, at scales of 10 m and larger, a variance larger than one is uncommon. Therefore, it is concluded that there exists a smallest possible scale for use of hydraulic pumping test results for estimating the effective hydraulic conductivity at scales typical for regional flow.