Continuous monitoring of health state is important in elder and chronically ill (but otherwise not critical) patients. Ideally, the patient should not be aware of being continuously monitored. This study describes an automatic and non-conscious heart rate monitoring system based on the ballistocardiogram (BCG). The bi-axial BCG can be recorded by two piezoelectric force sensors and high-gain amplifiers. Digital signal preprocessing has been implemented to increase the SNR and an algorithm to automatically detect the heart rate has been evaluated. The high gain required makes the system susceptible to forces other than those related to the cardiovascular system, such as those due to body movements. The sensitivity and positive predictive values obtained are 97.8 % and 98.4 % but if the records where the BCG is masked are discarded, the respective values increase to 99.2 % and 99.5 %. These values are comparable to those typically obtained for the ECG.