Human interactions are replete with emotional exchanges, and hence, the ability to decode others’ emotional expressions is of great importance. The present research distinguishes between the emotional signal (the intended emotion) and noise (perception of secondary emotions) in social emotion perception and investigates whether these predict the quality of social interactions. In three studies, participants completed laboratory-based assessments of emotion recognition ability and later reported their perceptions of naturally occurring social interactions. Overall, noise perception in the recognition task was associated with perceiving more negative emotions in others and perceiving interactions more negatively. Conversely, signal perception of facial emotion expressions was associated with higher quality in social interactions. These effects were moderated by relationship closeness in Greece but not in Germany. These findings suggest that emotion recognition as assessed in the laboratory is a valid predictor of social interaction quality. Thus, emotion recognition generalizes from the laboratory to everyday life.