T he patient was a 42-year-old man who presented 1 hour after ingestion of 25 tablets of 75 mg amitriptyline and an unknown amount of alcohol. He had a history of alcoholism and depression. He was unconscious, flaccid, and unresponsive, with grunting shallow breathing and absent gag reflex. The temperature was 35°C, heart rate was 120 beats/ min, and blood pressure was 140/90 mm Hg. On 14 L/min mask oxygen, his arterial pH was 7.29, the PAO 2 was 57, and the PACO 2 was 51 mm Hg. Urine toxicologic screen was ϩϩ for amitriptyline, and the blood alcohol level was 181 mg/dL. The rhythm strips recorded at the time are shown in Fig 1. He was intubated and ventilated. The routine laboratory tests and the chest x-ray film were unremarkable. The only specific overdose-directed therapy was MgSO4 20 mmol intravenously and forced diuresis; physostigmine and bicarbonate were withheld. Inotropes were not required, although pancuronium was used at induction. A 12-lead electrocardiogram (ECG) (Fig 2) was obtained 30 minutes after intubation.The obvious question in Fig 1 is whether there is an ST/T wave alternans. By definition, electrical alternans requires fixed rate and conduction. In Fig 1, the tachycardia is regular, with clear-cut identical P waves, with identical PR intervals, preceding the ventricular complexes. The latter, however, alternate in their terminal portion, the S wave following the almost embryonic R wave. Somewhat unexpectedly, it is the broader S waves that are associated with smaller subsequent repolarization deflections. Thus, sensu stricto, the appearance is more in keeping with alternating ventricular conduction delay than with electrical alternans. This conduction delay is of sufficient magnitude (well Ͼ.16 seconds) to predict the most severe complications of tricyclic overdose: seizures or ventricular tachyarrhythmias. 1 The full ECG was secured only when the patient was improving, with corresponding reduction in the QRS complex duration (Fig 2). It shows some reported features of tricyclic toxicity: terminal QRS axis greater than ϩ120°in the frontal leads 2 (but not the R/S ratio Ͼ.7 in aVR), 3 persistent conduction delay, now just over .12 sec, QT interval prolongation, and ST segment elevation in anteroseptal leads mimicking myocardial infarction 4 or the Brugada pattern. 5 This time, there is a true T-wave alternans, best seen in the V 1 rhythm strip: Both the QRS duration and the ST segment elevation remain constant. This leads one to surmise true alternans superimposed on alternating conduction delay in the Fig 1 rhythm strips. To our knowledge, T-wave alternans has not been reported in a tricyclic overdose.The alternans may not have independent prognostic value in this setting; the outlook may be subsumed in the associated QRS prolongation. 6 The latter, itself, may a function of the heart rate, 7 although not obviously so in this case: The rate remained fairly the same between the admission rhythm strips and the 12-lead ECG. The most important therapeutic factor normalizing the QRS complex must...