The susceptibilities of 106 strains of Aeromonas salmonicida to oxolinic acid (OXA), flumequine (FLU) and enrofloxacin (ENR) were established in two independent laboratories, using the M42-A disc diffusion protocols of the Clinical and Laboratory Standards Institute. The zone sizes obtained for OXA and FLU by both laboratories showed a bi-modal distribution. For OX-A there was clear separation of the two modal classes but for FLU the zones for the two classes showed a degree of overlap. In contrast, there was no clear evidence of bi-modality in the data obtained for ENR.Normalised resistance interpretation (NRI) was used to calculate laboratory-specific epidemiological cut-off values for wild type (WT) strains for all three agents. For OXA these cut-off values were 30 mm and 33 mm for the two laboratories and their application resulted in a complete concordance in the classification of the strains by the two laboratories. For FLU the NRI cut-off values were 39 mm and 40 mm and these resulted in an agreement between the laboratories in the classification of 95% of the 106 strains. The NRI cut-off values for ENR were 36 in and 40 mm and resulted in an agreement with the classification of 90% of these strains. When the zone sizes for the two modal classes overlap some strains with non wild type (NWT) phenotypes would generate zones larger than the WT cut-off value set by NRI analysis. Thus, definitely for ENR and possibly for FLU, the use of NRI cut-off values as provisional breakpoints could lead to some NWT strains being reported as sensitive.Analysis of the data obtained in this work demonstrated that the frequency of cross-resistance between the three agents was high and this is consistent with other reports in the literature. This high level of cross resistance provides a method by which the problems with the cut-off values set by NRI analysis can be avoided. It is argued that the use of OXA as a reporter agent represent an effective method of reducing error in reporting susceptibility to ENR and FLU. In this approach all strains shown to be NWT with respect to OXA would also be reported as NWT with respect to the other quinolones. The validity of this approach is dependent on there being a very low frequency of strains that are resistant to ENR and FLU but sensitive to OXA.It is recommended that, at the present state of our knowledge, the most effective method of reducing errors in determining the susceptibility of strains of A. salmonicida to quinolone agents is to employ OXA disc data as a proxy measure for susceptibility to the other agents in this class. 0 2007 Elsevier B.V. All rights reserved.