Objective: The number of police interventions with people presenting a mental health problem has been increasing during the past 30 years, and police services are becoming increasingly aware of the human resources and skills these interventions require. Our study addresses the characteristics explaining police time used and outcomes of interventions as police officers interact with people with mental illness.

Method: Using a police service administrative database from a large Canadian city, and an identification algorithm method, police interventions with people with mental illness were identified on 3 randomly selected days in 1 year. A content analysis of intervention logs was carried out to identify characteristics of those interventions: the call initiator, the location, and the final outcome of the intervention.

Results: Interventions with people with mental illness represent a small proportion (3%; n = 272) of all police interventions (n = 8485). General linear models show that the type of outcome is the most important factor in estimating the time required by police interventions. Arrests and hospitalizations are the least time-efficient outcomes, consuming 2.0 and 3.2 times, respectively, more time than informal dispositions. A multiple correspondence analysis shows that police interventions can be depicted in 2 dimensions, representing their main roles concerning people with mental illness, namely, to ensure the public safety and to protect the most vulnerable citizens. The more these services are required, the more police time will be required.

Conclusion: Education and partnerships between police services and mental health services are essential to a proper management of outcomes.