Plasma oxolinic acid (OXA) concentrations were measured in fish from a cage of farmed rainbow trout (Oncorhynchus mykiss) 1 day after the termination of medication. The fish were experiencing significant mortalities and following a diagnosis of vibriosis, OXA had been orally administered at 50 mg/kg for 6 days over a 9-day period. Samples from healthy fish (n=20), moribund (n=26) and dead fish (n=10) were analysed by HPLC. There was a dramatic difference in the OXA concentrations between healthy and moribund fish. In the moribund group, none of which showed signs of recent feeding, 85% of the fish had OXA levels below the LOQ (0.005 mg/l). In contrast, 95% of the healthy fish had OXA concentrations >0.015 mg/l and the mean OXA concentration (±standard deviation) was 0.156±0.152 mg/l. The mean OXA concentrations detected in the healthy fish in the farm were similar to those achieved in 30 laboratory held rainbow trout (O. mykiss) following the administration of OXA under similar conditions of salinity, temperature and dosing regimen. In these laboratory held fish, the mean plasma OXA concentration was 0.133±0.068 mg/l. The major difference between the distributions of OXA concentrations in the farm and laboratory populations was in the extent of fish to fish variation observed. In the healthy farmed fish, the percentage coefficient of variation (%CV) was 97% compared to a %CV of 51% in the laboratory held fish. The patterns of the daily mortality in the farmed population were analysed from 20 days before the initiation of therapy to 20 days after its completion but these data failed to provide unambiguous evidence as to the success or other wise of the therapeutic intervention. The moribund fish examined at the end of the therapy showed signs of systemic disease but from the majority, no bacteria were isolated. Strains of Vibrio anguillarum were isolated from some dead and moribund fish and these had a MIC of 0.0625 mg/l. It was, therefore, not possible to use the data generated in this work on the OXA plasma concentrations, the efficacy of the therapy and the MIC values of the infecting bacteria, to investigate the validity of any formula for estimating breakpoint MIC values.