To improve influenza vaccination rates, public health officials must understand potential factors in an individual’s decision to get a flu shot. Evidence suggests that an individual’s perception of risks related to infection and to vaccination may be influential, though surveys assessing these can be costly and slow. This paper triangulates data from multiple sources to study perceived risk in the context of flu vaccination. We create machine learning classifiers for relevant information on Twitter and statistically analyze discussions of perceived risk. We then compare Twitter discussions to data from a survey on perceived risk, showing how these data sources agree qualitatively. Future work will quantitatively examine these perceptions and potential misperceptions of risk.