Numerous instruments are commercially available to measure heart rate variability, yet little is known regarding the agreement between such instruments. The objective of this study is to assess agreement between measures of heart rate variability in three commercially available instruments. Thirty subjects (20 males) of median age 27.5 (range 19-59 years) underwent simultaneous ECG recordings, under three different resting conditions: supine, standing and supine with controlled breathing, using three commercially available analysers. Intraclass correlation coefficients tended to show excellent agreement (lower 95% C.I., R > 0.75) between all instruments under all conditions. However, further analysis of selected measurements using the limits of agreement method revealed large variation in values generated by all instruments. There was also an evidence of systematic bias between one instrument and the remaining two. The latter finding was due to discrepant ECG recording protocols that were unrelated to consistent operator timing. This study demonstrates that measures of HRV generated by the three instruments did not agree well in all cases. Discrepancies were due to the recording protocols of the systems. This may lead to incomparable results between instruments. It is therefore recommended that: (a) if different instruments are used in the same study or (b) multi-centre study designs are planned or (c) heart rate variability results are discussed with reference to studies using other instruments, levels of agreement need to be reported to ensure comparability.