Concern has been growing about a cyclical pattern or ‘periodicity’ in mental health care over the past 300 years, in which large-scale policy decisions have been made, and then reversed, paying little attention to the particular needs of differing patient groups. Turner (1985) drew attention to the reckless course followed by psychiatry, and how the “discipline that has lost its way” (Lancet, 1985) may have done so owing to a poor grasp of its past history. In the movement towards community care, ‘difficult’ or the so-called ‘unrewarding’ (to use Scott's (1970) term) patients are a particularly vulnerable group (Aviram & Segal, 1973; Bachrach, 1980). Increasingly, community services are compelled to cater for less disturbed patients through being selective about whom they will accept. To a degree, forensic services on both sides of the Atlantic have partially filled the gap so created in service provision. For this and other reasons, it is difficult to make a general statement about the effectiveness of deinstitutionalisation and community care. To be more specific, there are basic conceptual difficulties in defining continuity of care and identifying subgroups for which it has different implications (Bachrach, 1981), and there are major methodological difficulties impeding an analysis of such a diffuse and fragmented provision. Scull's (1984) description of the overnight abolition of state facilities for juvenile offenders in Massachusetts in the US between 1969 and 1973 points to the impossibility of making a comparison ‘before and after’, when so much dissolution and decentralisation of services has occurred (see also Jones, 1982). A few model alternatives were described soon after the closures, and Bachrach (1980) has written about the dangers of generalising from model programmes, highlighting the differences between demonstration efforts and reality. It is proposed here that the difficult and offender population might act as a ‘litmus paper’ for the efficiency, or otherwise, of community care in general.