Background: QT dispersion is considered to reflect inhomogeneity of myocardial repolarization. Method: The circadian variation of QT interval dispersion was examined in 95 healthy subjects using 24-hour Holter monitoring. Three different methods of lead selection were applied: all 12 leads (QTdisp 12), only precordial leads (QTdisp 6), and the pair of leads selected at 3 a.m. in which the longest and shortest QT intervals were found in each individual subject (QTdisp 2). Results: A preliminary methodological study including measurements from every minute in 10 subjects revealed no significant circadian variation using mean values of QTdisp 12, QTdisp 6, or QTdisp 2 obtained every hour, every 2, or every 4 hours, except in QTdisp 6, which demonstrated a significant circadian variation (P < 0.01) in 1-hour measurements. Analysis of all 95 subjects using measurements obtained every 4 hours revealed a significant circadian variation in QTdisp 12 and QTdisp 6 (P < 0.0001), whereas no circadian variation was seen in QTdisp 2. A subdivision into 10-year age groups revealed that subjects at age >50 years had a significant circadian variation in QTdisp 12 and QTdisp 6, but not in QTdisp 2. Only in males a significant circadian variation was seen in QTdisp 12 (P < 0.0001), whereas QTdisp 6 demonstrated a circadian variation both in females (P < 0.001) and in males (P < 0.0001). Conclusions: Selection of leads is of crucial importance for repetitive measurements of QT dispersion. Circadian variation was detected in subjects over 50 years of age, when all 12 or only the 6 precordial leads were taken into account.