Abstract: Techniques are provided for detecting the circadian state of a patient using an implantable medical device based on selected blood carbon dioxide (CO2) parameters. In one example, the implantable device tracks changes in end tidal CO2 (etCO2) levels and changes in maximum variations of pCO2 levels per breathing cycle (&Dgr;cycleCO2) over the course of the day and determines the circadian state based thereon. It has been found that average etCO2 levels are generally highest and average &Dgr;cycleCO2 levels are generally lowest while a patient is asleep and opposite while a patient is awake. Hence, by tracking changes in average etCO2 and &Dgr;cycleCO2 levels over the course of the day, circadian states can be detected. Minute ventilation and activity levels are used to assist in the determination of the circadian state. Additional techniques are directed to detecting the stage of sleep.