Social media can increase social capital, provide entertainment, and enable meaningful discourse. However, threats to safety experienced on social media platforms can inhibt users’ ability to gain these benefits. Threats to safety—whether real or perceived—detract from the pleasure people get out of their online interactions and damage the quality of online social spaces. While prior work has individually explored specific threats to safety – privacy, security, harassment – in this work we more broadly capture and characterize the full breadth of day-to-day experiences that influence users’ overall perceptions of safety on social media. We explore these perceptions through a three-week diary study (n=39). We contribute a novel, multidimensional taxonomy of how social media users define ’safety’, centered around security, privacy, and community. We conclude with a discussion of how safety perceptions can be used as a metric for social media quality, and detail the potential for enhancing safety perception through communityenhancing affordances and algorithmic transparency.