Liaison mental health services (LMHS) developed originally to address the mental health needs of people with physical illnesses in general hospitals and more recently to work also with people with mental health problems presenting at non-mental health services. The purpose of the present paper was to review empirical research on the structure, process and outcome of liaison mental health services using systematic review methods. Following a comprehensive search strategy, the authors reviewed 48 papers published between 1975 and 2001. There is an extensive international literature on LMHS, much of which describes the structure and process of liaison work. Studies evaluating the outcomes of liaison mental health services are fewer, and handicapped by methodological flaws, some of which are serious enough to cast doubts on the reported results. Professionals and clients value LMHS. LMHS based in accident and emergency (A & E) departments appear to ease the burden of general A & E staff, help clients access mental health services and reduce re-admission rates of people with mental health problems. There is little evidence supporting one model of configuring LMHS over another.