Rater reliability was evaluated for the system most widely used to assess copy and recall of the Rey Complex Figure: the Osterrieth (1944) 18-item scoring system. The study sample consisted of 95 subjects (49 males, 46 females), most of whom were elderly individuals (M = 59.83, SD = 15.21 years) suffering from memory impairment. Four raters rated copy and delayed-recall protocols, and three raters re-rated the protocols after an interval of 3 months. Results revealed excellent inter- and intra-rater reliability coefficients (.85-.97) for total scores. However, reliabilities for the 18 individual items ranged from poor (.14) to excellent (.96). Differences in both reliability and level of subject performance were observed as a function of item and conditions of copy versus recall. It is concluded that the Osterrieth scoring system supports excellent reliability in use with memory-impaired patients using total scores. Nevertheless, individual-item reliability would benefit from enhancement, for example, via amplified delineation of relevant decision criteria.