Severe premature babies present a risk profile higher than the normal population. Reasons are related to the incomplete development of physiological systems that support baby's life. Heart Rate Variability (HRV) analysis can help the identification of distress conditions as it is sensitive to Autonomic Nervous System (ANS) behavior. This paper presents results obtained in 35 babies with severe prematurity, in quiet and active sleep and in prone and supine position. HRV was analyzed in time and frequency domain and with nonlinear parameters. The novelty of this approach lies in the combined use of parameters generally adopted in fetal monitoring and "adult" indices. Results show that most parameters succeed in classifying different experimental conditions. This is very promising as our final objective is to identify a set of parameters that could be the basis for a risk classifier to improve the care path of premature population.