Short-term assessment of heart rate variability (HRV) is a non-invasive technique to examine ANS function. Within the literature, HRV is commonly referred to as a reliable measurement technique. The aim of this review was to assess the accuracy of this description based upon a comprehensive review of the available data concerning reliability of short-term HRV measures. Reviewing only studies using appropriate statistical analyses, it was determined that reliability coefficients for HRV measures were highly varied. Coefficients of variation ranged from <1% to >100%. Similar variation was found in studies using the intraclass correlation coefficient values, and limits of agreement. Reliability coefficients reported displayed some distinct patterns. Firstly, where measurements were made during interventions such as tilt or pharmacological stimulation, reliability was poorer than when HRV was measured at rest. Secondly, clinical populations displayed poorer reliability than healthy subjects. There was little effect of test–retest duration on reliability and although no single HRV measurement appeared less reliable than another, there was evidence that optimal data collection conditions for specific frequency domain measures exist. Describing HRV as a reliable measurement technique appears to be a gross oversimplification, as results of reliability studies are heterogenous, and dependent on a number of factors. Further studies are required, particularly in clinical populations to assess HRV reliability. Authors should refer to coefficients from similar populations measured under similar conditions when making future sample size calculations.