The meaningful sharing and combining of clinical results from different centers in the world performing boron neutron capture therapy (BNCT) requires improved precision in dose specification between programs. To this end absorbed dose normalizations were performed for the European clinical centers at the Joint Research Centre of the European Commission, Petten (The Netherlands), Nuclear Research Institute, Rez (Czech Republic), VTT, Espoo (Finland), and Studsvik, Nyköping (Sweden). Each European group prepared a treatment plan calculation that was bench-marked against Massachusetts Institute of Technology (MIT) dosimetry performed in a large, water-filled phantom to uniformly evaluate dose specifications with an estimated precision of +/-2%-3%. These normalizations were compared with those derived from an earlier exchange between Brookhaven National Laboratory (BNL) and MIT in the USA. Neglecting the uncertainties related to biological weighting factors, large variations between calculated and measured dose are apparent that depend upon the 10B uptake in tissue. Assuming a boron concentration of 15 microg g(-1) in normal tissue, differences in the evaluated maximum dose to brain for the same nominal specification of 10 Gy(w) at the different facilities range between 7.6 and 13.2 Gy(w) in the trials using boronophenylalanine (BPA) as the boron delivery compound and between 8.9 and 11.1 Gy(w) in the two boron sulfhydryl (BSH) studies. Most notably, the value for the same specified dose of 10 Gy(w) determined at the different participating centers using BPA is significantly higher than at BNL by 32% (MIT), 43% (VTT), 49% (JRC), and 74% (Studsvik). Conversion of dose specification is now possible between all active participants and should be incorporated into future multi-center patient analyses.