The prolongation of the ventricular repolarization and proarrhythmic effects (Torsade de Pointes: TdP) of five reference antibiotics were compared in four in-vitro models. 1. Using the patch clamp technique on the human ether-a-gogo-related gene (HERG) current, the rank order for blockade of the HERG-current (IC(50)) was: sparfloxacin (44 microM)>telithromycin=moxifloxacin=erythromycin (+/-100 microM). 2. Assessing their effects on action potential duration (APD(90)) and incidence of early afterdepolarizations in isolated rabbit Purkinje fibers, the rank order was: sparfloxacin>moxifloxacin>telithromycin>erythromycin (prolongation of APD(90) at 100 microM: 83%, 48%, 33% and 17% from baseline compared to +5% with solvent, P<0.05, respectively). 3. Assessing the drug effects on the APD(60), triangulation, reverse use-dependency, and instability in isolated Langendorff-perfused rabbit hearts, the rank order was: moxifloxacin>erythromycin>sparfloxacin>telithromycin. 4. Assessing their torsadogenic potentials (scores of effects on QT-interval, peak of the T wave to end of T wave: T(p-e), T(p-e)/QT ratio, R wave on T wave (R on T) and TdP in isolated rabbit left ventricular wedge preparations, the rank order for their TdP risk score was: sparfloxacin>erythromycin>moxifloxacin>telithromycin. Additional experiments with grepafloxacin indicate that the rank order to detect grepafloxacin-induced long QT was the wedge preparation>the Purkinje fiber>HERG>the isolated heart, where the isolated heart was unable to detect grepafloxacin-induced APD prolongation. The present study demonstrates that the first three in-vitro models can be used to assess the ability of antibiotic compounds to delay ventricular repolarization. However, with respect to their known clinical effects on QT and TdP incidence, the wedge preparation appears to be more predictive and suitable for detecting torsadogenic action of antibiotics.