Over a 10-month period, 33 patients with tardive dyskinesia (TD) were evaluated with the Abnormal Involuntary Movement Scale (AIMS) simultaneously and independently by two experienced and two inexperienced raters. The experienced raters generally had higher levels of agreement and their scores were more consistent over time. It is concluded that experience with TD influences AIMS inter-rater reliability and that it is useful to differentiate TD movements into the dimensions of quality, frequency, and amplitude, dimensions not currently used in the AIMS. The usefulness and difficulty of developing more specific guidelines for AIMS ratings are discussed.