Four samples of soybean oil refining process—a crude oil, a refined oil, an acid oil, and a deodorizer distillate—representing water contents from 0.02 to 3% were submitted to 16 laboratories for analysis by the Karl Fischer method ISO/TC 34/SC11/N99 recommended by the International Organization for Standardization. The laboratories were asked to perform 2 sets of replicate analyses within 2 weeks. Data population was homogeneous at the 95% confidence level except for the acid oil sample where a sampling problem caused wider than expected variation between laboratories. However, overall results indicated interlaboratory variations of the order of 15% at the 0.3% moisture level, in line with the 10% coefficient of variation suggested by the American Oil Chemists' Society at the 0.5% level. The experience of several participants in practicing the method leads to the recommendation that the draft method be modified to include special care in handling heterogeneous and semisolid or solid samples, use of test sample sizes of 5-25 g, adoption of alternative weighed water standardization, use of chloroform-methanol (1 + 1) or (2 + 1) premixed solvent, and use of reagent dilution for low water samples. The method has been adopted official first action by AOAC.