To determine the fraction of variational activity that is correlated on a breath-to-breath basis from uncorrelated random fluctuations, we performed autocorrelation analysis in 33 normal subjects during resting breathing. A calibrated inductive plethysmograph was used to nonobtrusively record 700 breaths in each subject. The group mean autocorrelation coefficients at a lag of 1 breath for each of the three primary breath components, tidal volume (VT), inspiratory time (TI), and expiratory time (TE), were significantly different from zero (p < 0.001). The autocorrelation coefficients for VT, 0.295 +/- 0.148 (SD), and TE, 0.259 +/- 0.121, were greater than that for TI, 0.201 +/- 0.135 (p < 0.001 and p < 0.01, respectively). The autocorrelation coefficients for each breath component remained significant for approximately 3 consecutive breaths (p < 0.001), indicating the presence of "short-term memory." Cross-correlation analysis revealed significant interrelationships (p < 0.001) for all component irrespective of which component was leading or following, with the exception of the pairing of VT in the leading breath and TI in the subsequent breath. In conclusion, in resting healthy subjects breath components display considerable breath-to-breath variability that is not completely random in nature, but which, instead, has a significant fraction of structured correlated variational activity.