The aim of the present study was to investigate the QT-RR interval relationship in ambulatory ECG recordings with special emphasis on the physiological circumstances under which the QT-RR intervals follow a linear relation. Continuous ECG recordings make it possible to automatically measure QT duration in individual subjects under various physiological circumstances. However, identification of QT prolongation in Holter recordings is hampered by the rate dependence of QT duration. Comparison of QT duration and QT interval rate dependence between different individuals implies that the nature of the QT-RR relationship is defined in ambulatory ECG. Holter recordings were performed in healthy volunteers at baseline and after administration of dofetilide, a Class III antiarrhythmic drug. After dofetilide, beat-to-beat automated QT measurements on Holter tapes were compared with manually measured QT intervals on standard ECGs matched by time. The QT-RR relationship was analyzed at baseline in individual and group data during three different periods: 24-hour, daytime, and nighttime. Data were collected under steady-state or non-steady-state conditions of cycle length and fitted with various correction formulae. Our study demonstrated an excellent agreement between manually and automated measurements. The classic Bazett correction formula did not fit the QT-RR data points in individual or group data. When heart beats were selected for a steady rhythm during the preceding minute, QT-RR intervals fit a linear relationship during the day and night periods, but not during the 24-hour period in both individual and group data. In contrast, in the absence of beat selection, data fit a more complex curvilinear relationship irrespective of the period. Our study provides the basis for comparison of QT interval durations and QT-RR relationships between individuals and between groups of subjects.