The prevalence of use of seclusion and restraints in psychiatric treatment has varied dramatically among institutions, according to previous studies. We investigated the factors predicting overall and "heavy use" of restrictive measures and differences in the population-based rates of use of seclusion and restraints in three university psychiatric centres in Finland (Turku, Tampere and Oulu) using a retrospective chart review. The material comprised all civil admissions to the study hospitals of working-aged people during a period of 6 months in 1996. There were significant differences among the studied centres as to the population-based level of use of seclusion and restraints. Oulu used significantly less seclusion but had a significantly higher level of use of restraints than Turku and Tampere. The individual institutions best predicted the overall use of restrictive interventions, whereas previous commitments and involuntary legal status on admission were factors predicting "heavy use" of these measures. Our results suggest that the implementation and monitoring of restrictive measures could be further harmonized.