Impairments in social behavior are frequently found in moderate to severe traumatic brain injury (TBI) patients and are associated with an unfavorable outcome with regard to return to work and social reintegration. Neuropsychological tests measuring aspects of social cognition are thought to be sensitive to these problems. However, little is known about the effect of general cognitive problems on these tests, nor about their sensitivity to injury severity and frontal lesions. In the present study 28 chronic TBI patients with a moderate to severe TBI were assessed with tests for social cognition (emotion recognition, Theory of Mind, and empathy), and for general, non-social cognition (memory, mental speed, attention, and executive function). The patients performed significantly worse than healthy controls on all measures, with the highest effect size for the emotion recognition test, the Facial Expressions of Emotion-Stimuli and Tests (FEEST). Correlation analyses yielded no significant (partial) correlations between social and non-social cognition tests. Consequently, poor performance on social cognition tests was not due to general cognitive deficits. In addition, the emotion recognition test was the only measure that was significantly related to post-traumatic amnesia (PTA) duration, Glasgow Coma Scale (GCS) score, and the presence of prefrontal lesions. Hence, we conclude that social cognition tests are a valuable supplement to a standard neuropsychological examination, and we strongly recommend the incorporation of measurements of social cognition in clinical practice. Preferably, a broader range of social cognition tests would be applied, since our study demonstrated that each of the measures represents a unique aspect of social cognition, but if capacity is limited, at least a test for emotion recognition should be included.