The effects of paced breathing (PB) on the reproducibility of heart rate variability (HRV) measurements were examined in 55 male subjects (age range: 20–54 years). Spectral components of HRV were measured under a combination of two respiratory conditions (spontaneous and 4s PB) and two postures (standing and supine). The procedures were repeated 3 weeks after the first measurement. Log-transformed low-frequency (lnLF) and high-frequency (lnHF) components of HRV were calculated from a 205s electrocardiography (ECG) recording. The coefficients of interindividual variations of lnHF and lnLF (ca. 13–16%) and the intraindividual variations of the frequency components (5–6%) were not significantly affected by PB. The coefficients of intraindividual variation of heart rate, ln HF and ln LF did not correlate with age in either posture. Effect sizes of PB on the intraindividual variation ranged from −0.04 to 0.13. Although intraclass correlation coefficients (ICCs) were slightly improved by PB in some cases, the differences were negligible. The above results suggest that PB provides a limited improvement in the reproducibility of HRV measurements, and metronome-guided respiration is not necessarily required for HRV measurement if subjects are reminded to avoid irregular respiration.