In order to characterize the variability and correlation properties of spontaneous breathing in humans, the breathing pattern of 16 seated healthy subjects was studied during 40 min of quiet breathing using opto-electronic plethysmography, a contactless technology that measures total and compartmental chest wall volumes without interfering with the subjects breathing. From these signals, tidal volume (VT), respiratory time (TTOT) and the other breathing pattern parameters were computed breath-by-breath together with the end-expiratory total and compartmental (pulmonary rib cage and abdomen) chest wall volume changes. The correlation properties of these variables were quantified by detrended fluctuation analysis, computing the scaling exponenta. VT, TTOT and the other breathing pattern variables showed α values between 0.60 (for minute ventilation) to 0.71 (for respiratory rate), all significantly lower than the ones obtained for end-expiratory volumes, that ranged between 1.05 (for rib cage) and 1.13 (for abdomen) with no significant differences between compartments. The much stronger long-range correlations of the end expiratory volumes were interpreted by a neuromechanical network model consisting of five neuron groups in the brain respiratory center coupled with the mechanical properties of the respiratory system modeled as a simple Kelvin body. The model-based α for VT is 0.57, similar to the experimental data. While the α for TTOT was slightly lower than the experimental values, the model correctly predicted α for end-expiratory lung volumes (1.045). In conclusion, we propose that the correlations in the timing and amplitude of the physiological variables originate from the brain with the exception of end-expiratory lung volume, which shows the strongest correlations largely due to the contribution of the viscoelastic properties of the tissues. This cycle-by-cycle variability may have a significant impact on the functioning of adherent cells in the respiratory system.