Background: Respiratory sinus arrhythmia (RSA) is associated with better health in children. Aim: The study was conducted to analyze the trajectory of RSA in 10-year-olds. Methods: A follow-up study on 120 healthy children (62 boys) aged 10.7 ± 0.5 years consisted of a standard 12-lead electrocardiogram, measurements of height, weight and blood pressure. The protocol was repeated after 3 years. Assessment of RSA based on semi-automatic measurements of RR intervals included: the difference between the longest and shortest RR interval duration (pvRSA), the root mean square of differences between successive RR intervals (RMSSD), the standard deviation of the RR interval length (SDNN) and their equivalents corrected for heart rate (RMSSDc and SDNNc). Results: A the first visit 61.7% of children presented with RSA; 51.7% 3 years later. 23.3% of them had RSA only on the first examination; 13.3% only on the second one. The pvRSA, RMSSD, and SDNN measured in 2019 did not differ significantly from their 2016 equivalents (p > 0.05). The decline in RSA defined by RMSSD was noted in 52.5% of children and in 54.2% when defined by SDNN. The corrected values decreased in 68.3 and 64.2% of the participants for RMSSDc and SDNNc, respectively. The students with RSA at both visits had lower heart rate (p < 0.001) and systolic blood pressure (p = 0.010) compared to those with rhythmic electrocardiograms. Conclusions: RSA in children is changeable, though its measurable indices should be adjusted to heart rate.