On account of the limited effectiveness of pharmacological treatments in Alzheimer's disease (AD), there is a growing interest on nonpharmacological treatments, including musical intervention. Despite the large number of studies showing the multiple benefits of music on behavioral, emotional and cognitive disorders of patients with AD, only a few of them used a rigorous method. Finally, the specificity of musical as compared to non-musical and pleasant interventions has rarely been addressed. To investigate this issue, two randomized controlled trials were conducted contrasting the effects of musical to painting (Study 1) or cooking (Study 2) interventions on emotional state of 33 patients with AD. The patients' emotional state was assessed by analyzing professional caregivers' judgments of the patient's mood, then facial expressions and valence of the discourse from short-filmed interviews. In the first study (n=22), each intervention lasted 3 weeks (two sessions per week) and the patients' emotional state was assessed before, during and after intervention periods. After the interventions, the results showed that facial expression, discourse content and mood assessment improved (more positive than negative expressions) as compared to pre-intervention assessment. However, musical intervention was more effective and had longer effects as compared with painting. In the second study (n=11), we further examined long lasting effects of music as compared to cooking by adding evaluation of the patients' emotional state 2 and 4 weeks after the last intervention. Again, music was more effective to improve the emotional state. Music had positive effects that remained significant up to 4 weeks after the intervention, while cooking only produced short-term effect on mood. In both studies, benefits were significant in more than 80% of patients. Taken together, these findings show that music intervention has specific effects on patients' emotional well being, offering promising methods to improve the quality of life of patients with AD.