We report on the first investigation of large-scale temporal associations between emotions expressed in online news media and those expressed on social media (Twitter). This issue has received little attention in previous research, although the study of emotions expressed on social media has bloomed owing to its importance in the study of mental health at the population level. Relying on automatically emotion-coded data from almost 1 million online news articles on disease and the coronavirus and more than 6 million tweets, we examined such associations. We found that prior changes in generic emotional categories (positive and negative emotions) in the news on the topic of disease were associated with lagged changes in these categories in tweets. Discrete negative emotions did not robustly feature this pattern. Emotional categories coded in online news stories on the coronavirus generally featured weaker and more disparate lagged associations with emotional categories coded in subsequent tweets.