Sixteen severely depressed patients participated in a double-blind randomized, crossover study to compare the effects of desmethylimipramine and imipramine on left ventricular function and the electrocardiogram. Following a drug-free week, patients had 3 weeks of therapy each with desmethylimipramine and imipramine. During each treatment period systolic time intervals, echocardiograms and high-fidelity electrocardiograms were recorded. There was no difference between desmethylimipramine and imipramine on (1) systolic time intervals, (2) shortening fraction or mean velocity of circumferential shortening, or (3) the electrocardiogram. There was a difference between the drug-free period and desmethylimipramine or imipramine on the PEPc (P less than 0.05) and the PEP/LVET ratio (P less than 0.05); on the R-R (P less than 0.05), PR (P less than 0.05), QRS (P less than 0.05), and QTc (P less than 0.05) intervals; but no difference on the LVETc or shortening fraction or the mean velocity of circumferential shortening. Drugs such as desmethylimipramine and imipramine which prolong intraventricular conduction can probably be expected to prolong the PEP and PEP/LVET. For this reason systolic time intervals have limitations in assessing myocardial function and the echocardiogram more reliably estimates myocardial performance in patients receiving tricyclic antidepressants.