This study presents an innovative approach to detect drowsiness by using photoplethysmography signals which is easily acquirable with non-invasive techniques. Drowsiness detection based on biological signals is being employed in precautionary personal safety. Autonomous Nervous System (ANS) activity can be measured non-invasively from the Pulse Rate Variability (PRV) signal obtained from photoplethysmography signal (PPG), that comprises alterations during, relaxation, extreme fatigue and drowsiness episodes. Our hypothesis is that these variations manifest on PRV. In this work we develop an on-line detector of drowsiness based on PRV analysis. The databases have been collected with the aid of an external observer who decides upon each minute of the recordings as drowsy or awake, and constitutes our data base.