1. Cardiac vagal outflow is the major factor determining the magnitude of heart rate (HR) variability analysed by traditional time and frequency domain methods. New analysis techniques, such as fractal and complexity methods, have been developed to probe non-linear features in HR behaviour that may not be detectable by traditional methods. 2. We investigated the effects of vagal blockade (glycopyrrolate i.v. 5 microg kg-1 h-1 for 2 h, n = 8 vs. unmedicated control group, n = 8) and various breathing patterns (n = 12) on two non-linear measures of HR variability--detrended fluctuation analysis (DFA) and approximate entropy (ApEn)--in healthy male volunteers. 3. Glycopyrrolate decreased the mean (+/-SD) ApEn from 1.46 +/- 0.18 to 0.85 +/- 0.24 (P = 0.001 in comparison with the control group), and increased the short-term (alpha 1) and intermediate-term (alpha 2) fractal scaling exponents of DFA, alpha 1 from 0.96 +/- 0.19 to 1.43 +/- 0.29 (P = 0.003) and alpha 2 from 1.13 +/- 0.10 to 1.34 +/- 0.14 (P < 0.001). 4. Decrease in fixed respiration rate from 15 to 6 breaths min-1 increased alpha 1 from 0.83 +/- 0.25 to 1.18 +/- 0.27 (P < 0.001), but decreased alpha 2 from 0.88 +/- 0.09 to 0.45 +/- 0.17 (P < 0.001) and ApEn from 1.26 +/- 0.12 to 1.10 +/- 0.14 (P = 0.028). Rapid breathing (24 min-1) had no influence on these non-linear measures of HR variability. Hyperventilation (15 min-1, tidal volume increased voluntarily by 0.5 l) decreased alpha 1 from 0.83 +/- 0.25 to 0.66 +/- 0.28 (P = 0.002) but did not affect alpha 2 or ApEn. 5. To conclude, vagal blockade alters the fractal scaling properties of R-R intervals (alpha 1, alpha 2) and reduces the complexity (ApEn) of HR behaviour. Both the fractal and complexity measures of HR variability can also be influenced by changes in the breathing pattern.