PFAS are a category of so-called forever chemicals that remain in the environment for very long times, and can endanger people's health by causing cancer and other harms. As states respond to PFAS contamination in drinking water, it is important to understand where the risks are greatest. We present preliminary results from a project to use advanced statistical methods (machine-learning) to predict which community water systems in Tennessee are at the greatest risk for PFAS. We found that our methods provide useful predictions that can help identify at-risk communities and guide response by the state and local community water systems.