Background There are three commonly used mental health interventions associated with policing: liaison and diversion, street triage and having specialist staff embedded in police contact control rooms. Crisis intervention teams (CITs), already used in the USA, are now attracting wider interest, including in the UK. Investment in these interventions is growing, so it is important to have evidence of their effectiveness. Aim To conduct a systematic literature review to answer questions about effectiveness of police‐mental health service models for responding to people with mental disorder and suspected offending or public safety problems. Methods A population/participant, intervention/indicator, comparator/control, outcome (PICO) framework was developed and keywords used to locate research studies within 29 databases. The 3179 results returned were screened by two researchers against preset inclusion and exclusion criteria. This resulted in 23 studies being included from which were taken the study methodology and findings and on which a narrative synthesis was conducted. Results Twenty‐three studies could be included, overall showing a positive impact of the interventions considered, but no well‐designed randomised controlled trials to test this optimistic view rigorously. Conclusions Overall, rather than indicating that one approach is more effective than another, the review points to the need for a multi‐faceted approach within a structured and integrated model, such as the CIT model. This is generally not the current pattern of interventions, and policy makers, service commissioners and providers may wish to review future options. Copyright © 2017 John Wiley & Sons, Ltd.