Anaesthetically equipotent doses of lidocaine, bupivacaine and a new bupivacaine-like local anaesthetic agent, ropivacaine, were injected into the left anterior descending coronary artery of pentobarbital-anaesthetized pigs. The aim was to study the cardiotoxicity of ropivacaine in relation to the two other drugs. A random, crossover, dose response study design was used. The following doses of the drugs were administered: lidocaine (L): 1,2,4,8 and 16 mg, bupivacaine (B): 0.25, 0.5, 1,2 and 4 mg and ropivacaine (R): 0.33, 0.66 1.33, 2.66 and 5.33 mg. Systemic haemodynamics, left ventricular dP/dT and a 12-lead electrocardiogram were recorded continuously during the study period. The drugs depressed cardiac contractility in relation to their local anaesthetic potency on the isolated nerve-4:3:1 (B:R:L). The prolongation of the ECG QRS-interval was regarded as a measure of electrophysiologic toxicity. Comparable prolongation of the QRS-interval was recorded after 2 mg of bupivacaine, 4.5 mg of ropivacaine and 30 mg of lidocaine. Thus, the electrophysiological toxicity ratio was 15:6.7:1 (B:R:L). Provided local anaesthetic potency data can be extrapolated from the isolated nerve preparation to regional anaesthesia in humans, ropivacaine appears to provide a greater margin of safety than bupivacaine, if inadvertently injected into the venous circulation.