The purpose of this study was to determine the value of designated musical instruments used by the therapist to elicit responses from persons with Alzheimer's disease (AD) during group sessions. In Experiment 1, 15 individuals in the early and middle stages of AD echoed rhythm patterns played by the therapist via the djembe, claves, paddle drum, and maraca. Results indicated significance for the rhythm patterns used, p < .05, and the types of instruments used, p < .05. A significant difference occurred for interaction of instruments by rhythms, p < .001. Rhythmic accuracy was highest when the therapist presented rhythms on the djembe, followed by the paddle drum, maraca, and claves. The most accurate rhythm pattern was 8 eighth-notes, the second most accurate was 4 quarter-notes, and the third was 2 eighths followed by a quarter note, repeated. Ten individuals in the middle to later stages of AD participated in Experiment 2. The treatment conditions consisted of 6 musical instruments used by the therapists to elicit responses during movement, singing, and rhythm activities. Average participation at any level (singing and moving/playing, singing only, and moving/playing only) in the 6 treatment conditions was highest during a cappella singing (63%), followed by djembe (61%), keyboard (60%), guitar and djembe (57%), and then guitar (54%) and autoharp (54%). Average participation at any level in the four activities, without any consideration of instruments, was as follows: 83% for the rhythm activity, 51% for the movement activity, and 49% for the first as well as second singing activity. Results indicated significant differences for the treatment conditions, p < .001, and for types of activities, p < .05. Significant difference, p < .01, occurred for interaction of treatment condition by activity.