Antipsychotics are frequently associated with QTc interval prolongation, a proposed marker for vulnerability to fatal ventricular arrhythmias, e.g. torsade de pointes (TdP). Little has been published on this topic in relation to clozapine. The objectives of this review were to: (i) calculate the frequency of QTc interval prolongation, T-wave abnormalities, TdP, ventricular tachycardia/fibrillation and sudden unexplained death in patients treated with clozapine and thioridazine from clinical trial and post-marketing reports; (ii) to compare these data with published findings for haloperidol, risperidone, olanzapine, sertindole and ziprasidone; and (iii) to correlate these clinical data with results from preclinical tests presently considered to be of predictive value for a compound's potential to cause QTc interval prolongation and TdP. A review of the global Novartis databases for clozapine and thioridazine and a Medline/Internet search for information on these cardiac events and for preclinical effects on the human ether-a-go-go related gene channels, action potential duration, and QT interval changes produced by the selected antipsychotics were performed. The clozapine database (2.8 million patient-years spanning 27 years) demonstrated that at therapeutic doses all but three reports of QTc interval prolongation and both of TdP were confounded by relevant co-medication/comorbidity. The literature review revealed that all antipsychotics considered except clozapine induced TdP and/or QTc interval prolongation at therapeutic doses. Preclinical in vitro tests appear to overestimate the risk of clozapine, haloperidol and risperidone to prolong QTc interval in patients and underestimate such a risk for sertindole and ziprasidone. Extrapolation of in vitro results to clinical events requires qualified interpretation.