Background Research in recent years has demonstrated that the use of coercive measures such as seclusion and restraint differs very much between hospitals within a country. In 2015, a central register for all coercive measures in the German federal state of Baden-Wuerttemberg has been established for 32 hospitals treating involuntary patients. The objective of the present study was to identify factors that determine the differences between these hospitals. Methods Data on coercive measures and diagnoses from the central register in 2015–2017 were linked with structural data of the 32 hospitals and their supply areas. Results On average, coercive measures were applied in 6.7% of cases (SD = 2.8%; Min–Max = 0.35–12.0%). The proportion of affected cases was significantly correlated with the proportion of involuntary patients (r = .56), the proportion of cases with affective or neurotic, stress-related and somatoform disorders (r = -.42), number of hospital beds (r = .44), a sheltered home associated with the hospital (r = .43) and number of addiction counseling centers per 100,000 inhabitants in the service area (r = -.39). The final regression model only included the proportion of involuntary cases as a significant predictor (standardized beta = 0.55, adjusted R2 = .27). Conclusions The predominating part of the considerable variance observed between hospitals could not be explained by structural variables. The proportion of involuntary patients had a significant impact, but a considerable amount of unexplained variance due to different practices within psychiatric hospitals remains.