An unconstrained method for the long-term monitoring of heart and breath rates during sleep is proposed. The system includes a sensor unit and a web-based network module. The sensor unit is set beneath a pillow to pick up the pressure variations from the head induced by inhalation/exhalation movements and heart pulsation during sleep. The measured pressure signal was digitized and transferred to a remote database server via the network module. A wavelet-based algorithm was employed to detect the heart and breath rates, as well as body movement, during sleep. The overall system was utilized for a total six-month trial operation delivered to a female subject. The profiles of the heart and breath rates on a beat-by-beat and daily basis were obtained. Movements during sleep were also estimated. The results show that the daily average percentage of undetectable periods (UPs) during 881.6 sleep hours over a 180 day period was 17.2%. A total of 89.2% of sleep hours had a UP of not more than 25%. The profile of the heart rate revealed a periodic property that corresponded to the female monthly menstrual cycle. Our system shows promise as a long-term unconstrained monitor for heart and breath rates, and for other physiological parameters related to the quality of sleep and the regularity of the menstrual cycle.