It is the responsibility of forensic psychiatric hospitals to detain and treat patients, gradually reintegrating them into society; decisions to release patients must balance risk to the public with maintaining the least restrictive environment for patients. Little is known about the factors considered when making such decisions and whether these factors have been empirically linked to future risk of violence. The current study explores the factors predictive of forensic treatment teams' recommendations for patients under the care of the Ontario Review Board (ORB). Factors differ depending on level of security; decisions on medium secure units were influenced by the presence of active symptoms and patients' overall violence risk level and decisions made on minimum secure units were influenced by the number of critical incidents that occurred within the recommendation year. Understanding the factors used to make recommendations to the ORB tribunal helps treatment teams to reflect on their own decision‐making practices. Furthermore, the results serve to inform us about factors that influence length of stay for forensic psychiatric patients. Copyright © 2016 John Wiley & Sons, Ltd.