A method is described for transferring the accuracy of values obtained by reference laboratories from one serum, which is used as a calibration standard, to another, which is treated as the unknown test sample. Six batches of lyophilised horse serum, of closely similar composition, were analysed by 10 European reference laboratories during 1978-1982, and used to obtain transfer values for one masterlot of a similar serum. No evidence was found that the transfer process introduced any inaccuracy. A comparison has been made, for bilirubin, creatinine, iron, phosphate, total protein and urea, between transferred values, the results obtained by direct analysis of the masterlot by reference laboratories, and consensus values found in two independent external quality assessment schemes. The transfer technique enables an evaluation to be made of these different procedures for assigning values to reference serum, and may reveal methodological differences, such as those described for iron and creatinine.