An experimental design was used to measure the effects of back massage on anxiety levels of elderly residents in a long-term care institution. Twenty-one residents, 17 females and four males, participated in the study. Subjects were randomly assigned to three groups which received 'back massage with normal conversation', 'conversation only' and 'no intervention' respectively. The dependent variable, anxiety, was measured prior to back massage, immediately following, and 10 minutes later, on four consecutive evenings. The Spielberger Self-Evaluation Questionnaire (STAI), electromyographic recordings, systolic blood pressure, diastolic blood pressure (DBP) and heart rate were used as measures of anxiety. Analysis of variance was used to examine differences in group mean scores over the pre-test to post-test, post-test to delayed time interval, and pre-test to delayed time intervals, Scheffé comparisons being made where indicated. With the exception of mean DBP which showed no change from pre-test to post-test and HR which increased from post-test to delayed time interval, there was a statistically insignificant decrease in mean scores on all variables in the back massage group from pre-test to post-test and from post-test to delayed time interval. There was a statistically significant difference in the mean anxiety (STAI) score between the back massage group and the no intervention group. The difference between the back massage group and the conversation only group approached statistical significance.(ABSTRACT TRUNCATED AT 250 WORDS)