Heart rate variability (HRV) is an important noninvasive method to quantify stress by measuring sympathetic and parasympathetic activity of the autonomic nervous system. Few studies exist on the repeatability and reliability of HRV in equids. The objectives of this study were to (1) compare the effect of different correction factors (CF) available in HRV analysis software on HRV indicator values and (2) to determine the repeatability and reliability of HRV indicators in an unrestricted (pasture) and a restricted movement (equine examination stocks) environment. Data were recorded on five consecutive days from six adult pony mares. Short-term tachograms were compared with regard to software CF by graphical and statistical (Friedman's and Wilcoxon signed rank test) comparison. The results showed that the specific CF influences the HRV indicator values. The strong CF was able to balance the elimination of artefacts without removing the variability of RR intervals and was subsequently used to determine repeatability and reliability. HRV indicators showed good repeatability over the 5 days using Friedman test (pasture, P = .162–.898; examination stocks, P = .29–.865), indicating that there were no significant differences between HRV indicator values. The reliability, represented by intraclass correlation coefficient (ICC) and coefficient of variation (CV), was poor-to-good for pasture data (ICC = 0.44–0.79; CV = 10–68.10) and examination stocks data (ICC = 0.22–0.83; CV = 16.07–62.29). Measurements obtained from the examination stocks were less reliable than those from pasture. Using normalized low-frequency and normalized high-frequency components improved reliability. Free-movement environment-based HRV recordings could ensure better reliability but may require the use of a stronger CF.