Time is an important aspect of all real world phenomena. In this paper, we present a temporal relations-based framework for discovering interesting patterns in smart environment datasets, and test this framework in the context of the CASAS smart environments project. Our use of temporal relations in the context of smart environment tasks is described and our methodology for mining such relations from raw sensor data is introduced. We demonstrate how the results are enhanced by identifying the number of individuals in an environment, and apply the resulting technologies to look for interesting patterns which play a vital role to predict activities and identify anomalies in a physical smart environment.