Objective-To investigate the diVerences in four formulae for heart rate correction of the QT interval in serial ECG recordings in healthy children undergoing a graded exercise test. Subjects-54 healthy children, median age 9.9 years (range 5.05-14.9 years), subjected to graded physical exercise (on a bicycle ergometer or treadmill) until heart rate reached > 85% of expected maximum for age. Design-ECG was recorded at baseline, at maximum exercise, and at one, two, four, and six minutes after exercise. For each stage, a 12 lead digital ECG was obtained and printed. In each ECG, QT and RR interval were measured (lead II), heart rate was calculated, and QTc values were obtained using the Bazett, Hodges, Fridericia, and Framingham formulae. A paired t test was used for comparison of QTc, QT, and RR interval at rest and peak exercise, and analysis of variance for all parameters for diVerent stages for each formula. Results-From peak exercise to two minutes recovery there was a delay in QT lengthening compared with RR lengthening, accounting for diVerences observed with the formulae after peak exercise. At peak exercise, the Bazett and Hodges formulae led to prolongation of QTc intervals (p < 0.001), while the Fridericia and Framingham formulae led to shortening of QTc intervals (p < 0.001) until four minutes of recovery. The Bazett QTc shortened significantly at one minute after peak exercise. Conclusions-The practical meaning of QT interval measurements depends on the correction formula used. In studies investigating repolarisation changes (for example, in the long QT syndromes, congenital heart defects, or in the evaluation of new drugs), the use of an ad hoc selected heart rate correction formula may bias the results in either direction. The Fridericia and Framingham QTc values at one minute recovery from exercise may be useful in the assessment of long QT syndromes. (Heart 2001;86:199-202) Keywords: paediatric exercise testing; QT interval; QTc formulae Since early in the 20th century physicians have known about the dependence of the QT interval on the heart rate, though its precise mechanism remains speculative. The first mathematical models describing the relation between the QT interval and heart rate were published as far back as 1920 by Bazett 1 and Fridericia 2 and were later converted into correction formulae.3 Attempts to describe the underlying QT interval independent of actual QT interval and heart rate resulted in the initial formulae. These formulae were repeatedly criticised, and many attempts have been made to propose more appropriate ones. [4][5][6][7][8][9][10] The shape relation between the QT and RR intervals is curvilinear. Mathematical forms used to model the relation between the QT interval and the RR interval include parabolic,