Open loop linear parametric modeling approach was applied to describe the variability of the ventricular depolarization and repolarization duration (i.e. the QT interval from the ECG). Several model structures were compared. The model maximizing the goodness of fit describes the QT interval as a linear combination of its own past values plus two exogenous influences (i.e. heart period interval and respiration) and a colored noise. When this model was applied to a protocol imposing a progressive increase of the sympathetic activity and modulation (i.e. the graded head-up tilt), the goodness of fit gradually decreased, thus suggesting a progressive uncoupling between QT duration and heart period that cannot be the result of influences unrelated to heart period changes.