The CIDRRI items were derived from interviews with 12 treatment supervisors in 4 Dutch forensic inpatient settings. The interviews focused on the patient characteristics the supervisors considered pertinent to the assessment of reoffending risk. In 1 of 3 studies, 15 cases were obtained from each of 5 participating hospitals; nearly all treatment supervisors and approximately half of all head nurses contributed to the study. Raters were instructed to perform paired ratings of each case within the same week, without discussing the checklist with one another. In the second study, nearly all treatment supervisors in three hospitals provided two or three double ratings each for patients of their own choosing. Respondents were instructed to complete the second rating 1 month after the first. The checklist for the second rating contained some additional questions on unusual circumstances that occurred between ratings that might have influenced scores. The third study compared ratings for all patients newly admitted to any of the eight hospitals during 1996 with those of all patients from seven hospitals who were about to re-enter the community on probationary leave between January 1, 1996, and December 31, 1998. In the absence of final recidivism data, predictive validity was approximated by comparing ratings for admittees with those for dischargees. Mean scores differed significantly and consistently in the expected direction, with newly admitted patients receiving more unfavorable ratings than patients beginning probationary leave. The findings warrant the conclusion that the CIDRRI is sufficiently reliable compared to available alternatives, and there are positive indications of its validity. 5 tables, 41 references, and appended CIDRRI