According to the World Health Organization, over 300,000 drowning deaths occur worldwide each year. One of the main contributors is lack of survival swimming knowledge; for example, the Redcross indicates that 50% of the population in the United States does not know how to swim. In this work, we explore the potential of using wearable technology to help address the lack of swimming training. Wearable technology like Fitbit or Apple Watch are smart devices that recognize people's activities and are increasingly integral to tracking one's health and well-being. When it comes to swimming or other aquatic activities, wearables have focused on professional and elite swimmers where devices support lap swimming strokes (backstroke, breaststroke, butterfly, freestyle) which are common in Olympic competitions. Our research expands lap swimming activity recognition by incorporating two survival swimming activities: treading water and sidestroke. We present a study which collects all six stroke types from a wrist-worn device, and we develop a machine learning algorithm that classifies activities with an F-measure of 0.94.