OBJECTIVES: To investigate the effects of hypoxia during sleep on linear and self-similar components of heart rate variability (HRV) in eight healthy subjects at high altitude on Mount Everest. METHODS: ECG was monitored by using an innovative textile-based device, the MagIC system. For each subject three night recordings were performed at sea level (SL), at 3500 m and 5400 m above SL. RR Interval (RRI) was derived on a beat-by-beat basis from the ECG and the VLF, LF and HF spectral components and the LF/HF ratio were estimated. Short- (alpha(1)) and long-term (alpha(2)) scale exponents as well as the recently proposed spectrum of self-similarity coefficients, alpha(n) were estimated by detrended fluctuation analysis (DFA). RESULTS: With respect to SL, all HRV parameters but one (alpha(2)) were significantly modified at 3500 m. However, at 5400 m they tended to return to the SL values and this was in contrast with the increase in the hypobaric hypoxia and in the number of central sleep apneas occurring at higher altitude. The only HRV index that displayed changes at 5400 m was the DFA alpha(n) spectrum, with alpha(n) values significantly lower than at SL for 20 < n <50 and higher for 200 < n <400, being n the box size. CONCLUSIONS: While the biological interpretation of these results is still in progress, our data indicates that the cardiac response to high altitude hypoxia during sleep can hardly be fully explored by traditional HRV estimators only, and requires the additional support of more sophisticated indexes exploring also nonlinear and fractal features of cardiac variability.