BACKGROUND: Dynamic instability in cardiac repolarization may contribute to drug-induced arrhythmogenesis. We hypothesized that intravenous cocaine would significantly destabilize repolarization as measured by QT variability. METHODS AND RESULTS: Twenty-nine cocaine-experienced volunteers not seeking treatment for cocaine addiction received randomized, sequential intravenous infusions of placebo or cocaine (20 and 40 mg). Five-minute epochs of digitized ECG were recorded 10 minutes before, during, and at intervals following the infusions. QT variability was measured using a semiautomated method and expressed as the log ratio of normalized QT variance to normalized heart rate variance (QTVI). Seventeen subjects received a repeat course of cocaine infusions 1 week later. Placebo infusion resulted in a small but significant increase in QTVI, while cocaine caused a highly significant, dose-dependent increase in QTVI that peaked at 10 minutes and dissipated by 45 minutes following infusion (P < 0.0001). The increase in QTVI was reproducible at 1 week (P = 0.8). CONCLUSIONS: Cocaine injection results in a significant dose-dependent increase in QT variability as indexed by QTVI. This destabilizing effect on repolarization may increase vulnerability to reentrant arrhythmias and may partially explain an increased risk of sudden cardiac death associated with cocaine use.