The purpose of this study was to examine the effect different cycling cadences have on heart rate variability (HRV) when exercising at constant power outputs. Sixteen males had ECG and respiratory measurements recorded at rest and during 8, 10 min periods of cycling at four different cadences (40, 60, 80 and 100 revs min−1) and two power outputs (0 W (unloaded) and 100 W (loaded)). The cycling periods were performed following a Latin square design. Spectral analyses of R–R intervals by fast Fourier transforms were used to quantify absolute frequency domain HRV indices (ms2) during the final 5 min of each bout, which were then log transformed using the natural logarithm (Ln). HRV indices of high frequency (HF) power were reduced when cadence was increased (during unloaded cycling (0 W) log transformed HF power decreased from a mean [SD] of 6.3 [1.4] Ln ms2 at 40 revs min−1 to 3.9 [1.3] Ln ms2 at 100 revs min−1). During loaded cycling (at 100 W), the low to high frequency (LF:HF) ratio formed a 'J' shaped curve as cadence increased from 40 revs min−1 (1.4 [0.4]) to 100 revs min−1 (1.9 [0.7]), but dipped below the 40 revs min−1 values during the 60 revs min−1 1.1 (0.3) and 80 revs min−1 1.2 (0.6) cadence conditions. Cardiac frequency (fC) and ventilatory variables were strongly correlated with frequency domain HRV indices (r = −0.80 to −0.95). It is concluded that HRV indices are influenced by both cycling cadence and power output; this is mediated by the fC and ventilatory changes that occur as cadence or exercise intensity is increased. Consequently, if HRV is assessed during exercise, both power output/exercise intensity and cadence should be standardized.