Heart rate variability (HRV) analysis is used to examine morbidity and mortality in premature neonates. We developed a system to acquire and analyze full frequency spectrum HRV based on ECG signal to identify frequency ranges associated with responses to "normal care" of the NICU neonate. The system will allow real-time monitoring of specific frequency ranges at the bedside. Twenty NICU newborns were enrolled under University of Miami IRB approved protocol. Infants were recorded before, during and after procedures associated with normal NICU care. ECG signals were sampled from analog output from a standard bedside monitor using a custom interface based in an Arduino system. Interface was connected to a computer running a LabView program for realtime process. Pam-Tompkins QRS detection algorithm was tuned to spectral characteristics of the preterm ECG signal; HRV computation was done using the Lomb-Scargle algorithm. ECG acquisition was safe, reliable and appropriate for HRV computation. 1186 minutes of ECG were sampled at 1000SPS and processed. The QRS detection algorithm was reliable, requiring minor intervention to eliminate artifacts. HRV computation using non-uniformly spaced samples was appropriate for ECG analysis. HRV can be used in real time to monitor neonatal response to normal care.