We describe a prototype context aware perioperative information system to capture and interpret data in an operating room of the future. The captured data is used to construct the context of the surgical procedure and detect medically significant events. Such events, and other state information, are used to automatically construct an electronic medical encounter record (EMR). The EMR records and correlates significant medical data and video streams with an inferred higher-level event model of the surgery. Information from sensors such as Radio Frequency Identification (RFID) tags provides basic context information including the presence of medical staff, devices, instruments and medication in the operating room (OR). Patient monitoring systems and sensors such as pulse oximeters and anesthesia machines provide continuous streams of physiological data. These low level data streams are processed to generate higher-level primitive events, such as a nurse entering the OR. A hierarchical knowledge-based event detection system correlates primitive events, patient data and workflow data to infer high-level events, such as the onset of anesthesia. The resulting EMR provides medical staff with a permanent record of the surgery that can be used for subsequent evaluation and training. The system can also be used to detect potentially significant errors. It seeks to automate some of the tasks done by nursing staff today that detracts from their ability to attend to the patient.