The interquartile range has become the international de facto standard for transfer pricing benchmarking studies. However, it is not a statistically robust method to trim a range of values. Instead of quartiles, confidence intervals of quartiles should be applied. This approach is in line with OECD TP guidance, and furthermore provides a clear criterion for an often neglected aspect of OECD TP guidance: that benchmark ranges should only be narrowed in the case of a sizeable number of observations. The author suggests that the full range should always be used for sets of less than 27 comparables. The suggested methodology can be easily implemented using standard spreadsheet software.