The modifiable areal unit problem (MAUP) is seldom accounted for in economic geography cluster studies. This issue is particularly problematic as clusters play such a central role in regional development policy. In this study, we tested for the appearance of the MAUP when examining workplace category clusters over three different administrative scales in Stockholm County, Sweden. The measures used were the standardised ratio, global autocorrelation, and local autocorrelation. We discovered not only that the appearance of clusters depended on the administrative unit scale but also that the location of the clusters, based on the local indicator of spatial association, fluctuated wildly over the various administrative scales. Thus, our conclusion was that the robustness and validity of workplace category clusters should always be challenged if the MAUP has not been taken into account, or at least tested for.