The concept of positive deviance (PD), which highlights uncommon practices that reduce risk in low-resource communities, has been effective in community mobilization and programming to improve health outcomes. We present a protocol for extending the concept to analysis of existing public health data. The protocol includes assessing whether PD fits the situation, identifying positive deviants, and identifying behaviors associated with positive deviants' healthy outcomes. Analyzing existing datasets from a PD perspective may aid public health nurses in efforts to reduce health disparities. The effectiveness of our protocol will be clarified in future research.