Repeat visitors to psychiatric emergency services constitute 7 to 18 percent of the total patients and account for up to a third of the visits. This frequently difficult-to-treat group has been described in six controlled and two uncontrolled studies. In a critical review of findings and methodology the authors interpret and compare the available data about repeaters. Their major conclusions are that repeaters are more likely to lack social supports, to be currently in psychiatric treatment, and to have a chronic illness. Research approaches to further elucidate these patients' characteristics are suggested.