Cross correlation is a mathematical function whereby spectral analysis is used to describe the relationship between heart-rate fluctuations (256 R-R intervals) and respiration (simultaneously obtained by pneumotacograph). To assess its usefulness for testing autonomic integrity, cross correlation and deep breathing were compared in 141 diabetic subjects (aged 39 +/- 14 yr) and in 77 control subjects (aged 33 +/- 13 yr). To characterize patients, Valsalva maneuver, 30:15 ratio, tilt, and handgrip tests were performed in 96 of these patients; 23 had two or more abnormal tests (group A), 28 had one (group B), and 45 had none (group C). Sensitivity to parasympathetic withdrawal was compared in 9 control subjects (aged 26 +/- 4 yr) by four sequential 0.01-mg/kg i.v. atropine administrations. Reproducibility was compared in 11 control subjects (aged 25 +/- 2 yr) by repeating the tests four times for 2 consecutive days. Considering all 141 patients, cross correlation and deep breathing were less than 2SD of the mean of control subjects in 64 and 36 subjects, respectively. Considering patients who also performed other tests of autonomic function, cross correlation and deep breathing were less than 2SD of the mean of controls in 42 and 30 subjects, respectively (group A, 20 and 15; group B, 12 and 9; group C, 10 and 6). Cross correlation had better reproducibility than deep breathing (C.V. 10.3 vs. 30.6% at 6 breaths/min) and greater sensitivity to atropine (after the 1st injection, cross correlation and deep breathing decreased to 34.6 and 48.2% of baseline values, respectively; P less than .05).(ABSTRACT TRUNCATED AT 250 WORDS)