It is nearly thirty years since the last inter-laboratory study was carried out for amino acid racemization (AAR) analysis using powdered fossil material (Wehmiller 1984). Since then there have been major changes in sample preparation and instrumentation, and it was considered timely to coordinate a new inter-laboratory study in support of current methodologies. In 2010, two such studies were undertaken. The first of these, coordinated by Wehmiller (this edition), used homogeneous hydrolysates of Pleistocene mollusc and eggshell materials and focused on the agreement of analytical measurements between laboratories, without interference from differing sample preparation procedures. The second (this study) was designed specifically as a proficiency test. Unlike previous inter-laboratory comparisons that have focussed on precision estimates, the purpose of this study was to carry out an evaluation of measurement bias by comparing the measurement results of laboratories carrying out their routine methods, including extraction, against the consensus values. Participants were sent one dried sample of a mixed amino acid standards solution and five homogeneous powders: two Pleistocene mollusc test materials prepared from material (ILC-A) supplied and used by Wehmiller in previous inter-laboratory studies (1984; and this edition), one Pleistocene opercula test material from the terrestrial gastropod, Bithynia tentaculata, and two heat-treated modern ostrich eggshell test materials. Results from this study demonstrate that whilst individual laboratory precision may be excellent, suggesting good control of random error influences (less than 1% for replicate measurements by some individual laboratories), agreement between methods, or even between laboratories carrying out the same method, may be very different. Trueness evaluation (determined as the relative percentage bias) reveals the extent of the disagreement reflected by the inter-laboratory variability. Individual laboratory D/L value biases of 10-30% or more when compared to the consensus values are not uncommon. We demonstrate why bias contributions should also be included in AAR uncertainty estimation and recommend that the preparation of defined reference materials are seen as a priority in order to control and correct for systematic error influences in the analytical system.