The first international round-robin exercise for the measurement of the long-lived radionuclide 10Be has been conducted. Ten participating accelerator mass spectrometry (AMS) facilities have each measured three samples at the 10−12 to 10−1410Be/9Be level. All results have been made traceable to the NIST SRM 4325 standard to avoid additional discrepancies that arise when different facilities use different calibration materials. Hence, the data concentrates on pure measurement distinctions. Multivariate statistical investigations have been performed to reveal a bias between facilities, i.e. two distinguished groups could be identified. Maximum discrepancies between two single facilities are in the range of 6–31% depending on the absolute 10Be/9Be value. These findings should be considered when comparing 10Be data produced at one AMS facility with that produced at another facility, which is e.g. often the case for in situ 10Be dating studies. Round-robin exercises are a very helpful tool as part of an overall quality assurance scheme to improve the accuracy, and not only the precision, of AMS data.