To investigate the incidence of coercive measures in standard psychiatric care in different psychiatric hospitals. We developed a common documentation of mechanical restraint, seclusion, and medication by coercion, and introduced it in 10 participating hospitals. We developed software able to process the data and to calculate four key indicators for routine clinical use. 9.5% of 36,690 cases treated in 2004 were exposed to coercive measures with the highest percentage among patients with organic psychiatric disorders (ICD-10 F0) (28.0%). Coercive measures were applied a mean 5.4 times per case and lasted a mean 9.7 h each. The incidence and duration of coercive measures varied highly between different diagnostic groups and different hospitals. Use of detailed guidelines for seclusion and restraint was associated with a lower incidence of coercive measures. Data interpretation should consider numerous confounding factors such as case mix and hospital characteristics. Suggestions on how to cope with ethical and technical problems in the processing of large multi-site data sets in routine clinical use are made.