Electroconvulsive therapy (ECT) is the most effective treatment for severe depression, and different forms are increasingly used in clinical practice. This study investigated the acute cardiac effects of different forms of ECT: bitemporal and bifrontal (1.5 times seizure threshold), and right unilateral (RUL) (five times seizure threshold). For RUL ECT, the effect of stimulus pulsewidth (1.0 or 0.3 ms) was also examined. Electrocardiograms recorded just prior to and during the ECT stimulus in 476 ECT treatments in 114 patients were examined. The degree of bradycardia (any slowing of heart rate) and incidence of asystole (absence of heart beats for ⩾5 s) during the ECT stimulus were measured from these traces. Regression analyses estimated the contribution of patient and ECT treatment factors to the risk of bradycardia and asystole. Bifrontal ECT was associated with less severe bradycardia than bitemporal or RUL ECT (p<0.001). Modelling showed, for a mean pre-ECT heart rate of 85 beats per minute (bpm), expected heart rates during the stimulus were 78 bpm (bifrontal), 46 bpm (bitemporal) and 35 bpm (RUL). Bifrontal ECT was also associated with a lower incidence of asystole than RUL ECT (corrected odds ratio 1:207) and bitemporal ECT (corrected odds ratio 1:24). Ultrabrief pulsewidth stimulation resulted in lesser bradycardia and asystole than standard pulsewidth stimulation for RUL ECT. Modelling showed, for a mean pre-ECT heart rate of 86 bpm, expected heart rates were 43 bpm (ultrabrief RUL) and 26 bpm (RUL). Bradycardia and asystole were relatively common side-effects during the ECT stimulus. Bifrontal ECT was associated with the lowest risk of bradycardia and asystole during ECT and should be considered for patients at risk of arrhythmias and prolonged asystole during ECT.