Inferring the significance of physiological events in the natural environment is difficult and requires using many sensors to capture context. Stress reactions can occur from both physical and mental stressors. To understand the factors involved in the stress reaction, a model must be developed to interpret an individual's reaction in context. Using multiple sensors and context data, we hope to develop algorithms to distinguish between mental stress, physical stress and a relaxed state. The results of these algorithms could be used to help make people aware of how daily tasks effect their stress level. Real-time stress detection algorithms could also be used to monitor driver stress and determine if it is advisable to interrupt the driver with cell phone calls or navigation assistance.