Background. The statement that past behaviour is the best predictor of future behaviour has empirical support in respect of whether an individual will, or will not, commit a criminal offence. People who have offended in the past are more likely to offend in the future. The aims of this study were to develop an instrument and to examine whether the same statement applies in respect of the nature and circumstances of successive offences committed by the same person.

Method. A rating instrument to describe seven variables relating to the nature and circumstances of a serious offence, the SODI, was developed by the authors. Inter-rater reliability was measured when the instrument was applied to 80 offences committed by 40 patients leaving high secure hospitals. The data were examined for evidence of similarity in the nature and circumstances of successive offences.

Results. For five of the seven items of the instrument the kappa coefficients for inter-rater reliability were >0·65. No significant associations, in terms of SODI ratings, were found between the offence that led to hospital admission and that which was committed after discharge.

Conclusions. The SODI is a reliable instrument for the description of serious offences committed by this group. Risk assessments in psychiatry should be informed by an awareness that in only a proportion of cases will the nature and circumstances of any serious re-offence resemble the nature and circumstances of the offence which contributed to a patient's admission to hospital.