We studied the seclusion records of an adolescent unit for a six-year period. Problems in the analysis of typical seclusion data are examined and statistical methods that overcame them are explained. Only a few relationships were found between the seclusion regime and available measures of patient and ward characteristics, though there was great variability in the frequency with which individuals were secluded and the duration of various episodes of seclusion. The average duration of seclusion (31 minutes) was much less than had been reported by other investigators. Furthermore, the durations became shorter over the period of study. The implications of these findings for further research and practice are discussed.