Considerable resources have been devoted to gathering data for the measurement of money market activity. However, little is known about the differences between available data and the structural effects of methodological choices. We use the novel dataset MMSR and compare it to data derived from a Furfine-type algorithm and survey data. The deviations in volumes and interest rates are driven by the asymmetric measurement of transactions, in particular affecting individual classes of banks, cross-border loans and specific types of loans. These differences are significant in terms of magnitude and affect overall rates and volumes. Even fundamental questions like the share of cross-border transactions depend on which data is used.