This study evaluated the systemic toxicity, arrhythmogenicity, and mode of death of convulsant and supraconvulsant doses of lidocaine, bupivacaine, and ropivacaine. Experiments in awake dogs were designed to mimic the clinical situation of an accidental intravenous (IV) injection of local anesthetics. On the first experimental day, lidocaine (8 mg.kg-1.min-1), bupivacaine (2 mg.kg-1.min-1), and ropivacaine (2 mg.kg-1.min- 1) were infused intravenously until seizures occurred (n = 6 for each group). The average dose and arterial plasma concentration at seizure onset was 20.8 +/- 4.0 mg/kg and 47.2 +/- 5.4 micrograms/mL for lidocaine, 4.31 +/- 0.36 mg/kg and 18.0 +/- 2.7 micrograms/mL for bupivacaine, and 4.88 +/- 0.47 mg/kg and 11.4 +/- 0.9 micrograms/mL for ropivacaine. The margin of safety between the convulsive and lethal doses was determined by administering two times the convulsive dose 24 h later. Two dogs given lidocaine died because of progressive hypotension, respiratory arrest, and finally cardiovascular collapse with an average peak plasma concentration (Cmax) of 469 micrograms/mL. No ventricular arrhythmias were observed in this group. Ventricular arrhythmias occurred in five of six dogs receiving bupivacaine. Four animals died because of hypotension, respiratory arrest, and cardiovascular collapse. One additional animal died because of ventricular fibrillation. The Cmax for bupivacaine was 70.1 +/- 14.6 micrograms/mL in nonsurvivors. In the ropivacaine group one animal died because of hypotension, respiratory arrest, and cardiovascular collapse (Cmax = 72.4 micrograms/mL). A surviving dog had transient premature ventricular contractions. Twenty-four hours later three times the convulsive dose was administered to the survivors.(ABSTRACT TRUNCATED AT 250 WORDS)