SummaryAn HPLC column-switching method has been developed and validated for the enantioselective determination of (R)-and (S)-carvedilol in human plasma. Sample preparation was performed either off-line, by extraction with trichloromethane and back-extraction into 0.01 M aqueous citric acid which was injected on to a LiChrosorb RP 8 column, or on-hne, by injecting diluted (0.1 M formic acid) plasma on to a LiChrosorb ADS column. In both instances separation was performed by gradient elution and on-line transfer of the fraction containing the carvedilol on to an enantioselective Teicoplanin column. The enantiomers of carvedilol were separated isocratically by use of methanol-acetonitrile-triethylammonium acetate, 70:30:0.05 (v! v/w), as mobile phase. With fluorescence detection the limits of quantitation were 0.30 ng mL 1 for (R)-carvedilol and 0.26 ng mL 1 for (S)-carvedilol; these were sufficient to enable investigation of the effect of exercise on plasma concentrations of (R)-and (S)-carvedilol after oral administration of either the racemate or the pure enantiomers.Although the operating conditions were optimized for sample preparation by on-line deproteination on a LiChrospher RP 18 ADS column, the complete method was insufficiently rugged for analysis of large numbers of plasma samples -the enantioselectivity of the Teicoplanin colu mn deteriorated too rapidly because of the transfer of ena ntioselectivity-poisoning interferences which could not be suppressed sufficiently. In contrast the liquid-liquid sample-extraction procedure combined with column switching resulted in a analytical method with long-term stabihty.