BACKGROUND An investigation to determine which prognostic variables are associated with behavioural treatment failure in obsessive-compulsive disorder (OCD). Empirically established prognostic variables measured at the start of treatment may lead to adjusted treatment programmes for these patients.

METHOD Forty patients, diagnosed with OCD, received a standardised treatment consisting of 18 sessions in vivo exposure and response prevention. Compulsive behaviour (MOCI) and obsessive fear (ADS) were the outcome measures. Prognostic variables included were initial severity of OC complaints, initial level of depression, problem duration, patients' motivation for treatment, quality of the therapeutic relationship, and marital dissatisfaction.

RESULTS Greater initial severity of complaints (P < 0.01), and depression (P = 0.03) predicted poorer outcome for compulsive behaviour. Greater initial severity of complaints (P < 0.01), and the conjoint variables higher level of depression, longer problem duration, poorer motivation for treatment, and dissatisfaction with the therapeutic relationship predicted poorer outcome for obsessive fear (P < 0.01).

CONCLUSIONS The complaint-related variables of initial severity, initial depression, and problem duration, and the non-specific treatment variables of patients' motivation and quality of the therapeutic relationship, affect behavioural treatment outcome in OCD.