Author's reply: I am grateful to Dr Pearce (2000, this issue) for drawing attention to two significant pieces of research. The study by Bateman and Fonagy (1999) was omitted simply because it had not been published at the time of writing. I agree that it is an important study, which demonstrates the effectiveness of an eclectic, but psychoanalytically based treatment approach. Moreover, the model of day treatment for BPD is one that may potentially be replicated in the National Health Service and is already attracting considerable interest.

The study by Lees et al (1999) is different in that it is essentially a meta-analysis of studies of the effectiveness of therapeutic communities, rather than the treatment of BPD. Furthermore, the conclusion that Pearce quotes refers to their effectiveness in the treatment of personality disorder in general. The analysis included studies of both concept-based therapeutic communities devoted to the treatment of substance misuse and forensic units where there is likely to be a preponderance of patients with antisocial personality disorder. Furthermore, a wide range of research methods and outcome measures was used in the studies included, many of them not specific for BPD. Although the findings of Lees et al are encouraging, their relevance to the treatment of BPD is therefore somewhat uncertain.

However, these studies do form part of an increasing and welcome trend towards the empirical evaluation of treatments for personality disorders. It is to be hoped that this trend will continue and will stimulate a debate about the rational provision of services for this important group of patients.