The issue of emotion recognition in real-life settings, lacking a clear criterion of the nature of the underlying emotion, is raised. After reporting their luggage lost, 110 airline passengers were asked to rate their emotional state (subjective feeling criterion). The agents who had processed the claims were asked to rate the passengers' emotional state (objective behavior criterion) as well as their own feelings. An excerpt of the videotaped interaction for 40 passengers was rated for emotional state by judges on the basis of (a) verbal and nonverbal cues or (b) nonverbal cues only. As predicted, the data show that judges' inferences in both exposure conditions correlate more strongly with the objective behavior (agent ratings) than the subjective feeling criterion (self-ratings). Using the Facial Action Coding System (FACS), objectively coded “felt” (but not false) smiles correlated positively with a good humor scale in both criteria and judges' ratings.