Relaxing music was hypothesized to buffer the general noise level typically found in dining rooms of nursing homes, exert a calming effect, and thus reduce agitated behaviors among residents with severe cognitive impairment. Twenty-nine nursing home residents with severe cognitive deficits participated in a 4-week protocol in which, following baseline observations (week 1), relaxing music was introduced (week 2), removed (week 3), and reintroduced (week 4). Subjects were observed in terms of total number of behaviors of the Cohen-Mansfield Agitation Inventory present during a given week, and the number of behaviors present on the subscales of aggressive, physically nonaggressive, verbally agitated, and hiding/hoarding behaviors. Significant reductions were observed on the cumulative incidence of total agitated behaviors (63.4%); as well as the cumulative incidence of physically nonaggressive behaviors (56.3%) and verbally agitated behaviors (74.5%). No significant reductions were noted in terms of aggressive behaviors and hiding/hoarding behaviors. Where significant reductions were achieved, a distinct pattern was observed. Agitation decreased during week 2, increased again during week 3, only to decrease again in week 4. In addition, variance effects were noted as well, as ranges and standard deviations of agitated behaviors narrowed over time. These findings are interpreted within Hall and Buckwalter's (1988) model of a progressively lowered stress threshold among dementia patients.