We present an energy-efficient method for Indoor/Outdoor detection on smartphones. The creation of an accurate environmental exposure detection method enables crucial advances to a number of health sciences, which seek to model patients’ environmental exposure. In a field trial, we collected data from multiple smartphone sensors, along with explicit indoor/outdoor labels entered by participants. Using this rich dataset, we evaluate multiple classification models, optimised for accuracy and low energy consumption. Using all sensors, we can achieve 99% classification accuracy. Using only a subset of energy-efficient sensors we achieve 92.91% accuracy. We systematically quantify how subsampling can be used as a trade-off for accuracy and energy consumption. Our work enables researchers to quantify environmental exposure using commodity smartphones.