Complexity of the RR interval variability decreases during sympathetic activation and vagal withdrawal. The QT interval depends largely on the previous RR, thus suggesting that QT variability complexity should follow the decrease of the RR variability complexity. We hypothesize that, despite the decrease of the RR variability irregularity during sympathetic activation, QT variability complexity remains high and this behavior is owing to the increase of the complexity of the QT variability unrelated to RR changes. We recorded beat-to-beat RR and QT variability in 15 healthy young subjects during graded head-up tilt (HUT) at 0, 15, 30, 45, 60, 75 and 90 degrees. Respiration (R) was acquired via thoracic belt and sampled once per cardiac beat. A supine resting period preceded each HUT session. A multivariate dynamical adjustment model described the QT as the linear combination of m past QTs, m past RRs, m past R samples plus an autoregressive noise (wQT). The model allows QT variability decomposition into partial processes due to RR (QT/RR), R (QT/R) and noise unrelated to RR and R dynamics (QT/wQT). Complexity of the RR, QT, QT/RR, QT/R and QT/wQT series was computed via sample entropy (embedding dimension=2; tolerance=0.2×standard deviation; series length=256). Association between sample entropy and HUT angles was assessed via Pearson correlation coefficient r and type I error probability p. The model order m was optimized via Akaike information criterion. The RR variability complexity decreased progressively with HUT angle (r=-0.598, p=1.59×10-11), while that of the QT dynamics remained unvaried (r=0.136, p=1.67×10-1). The complexity of the QT/wQT dynamics increased gradually with the magnitude of the orthostatic challenge (r=0.199, p=4.18×10-2), while that of QT/RR and QT/R was unrelated to it (r=-0.160, p=1.02×10-1 and r=0.021, p=8.35×10-1 respectively). During sympathetic activation the increase of the QT/wQT complexity offsets the tendency to decrease of the QT/RR irregularity.