This is a cluster randomised controlled trial composed of two research arms: exercise training group (experimental group) and a social activity group (control group).

The recruitment process will be made only in NHs (Nursing Home) where the staff will have voluntarily accepted to participate in the study. We estimated to recruit 140 PWD (person with dementia) from around six-to-eight NHs (i.e., in average a minimum of 18 and a maximum of 24 PWD per NH). NHs constitute the unit of randomisation; NHs will be randomised to study groups using a 1:1 ratio (it means that all the participants living in a particular NH will participate in either the exercise training or the social activity). The randomisation will be stratified by the prevalence of dementia among NH residents; this stratification will provide homogeneity across research's arms, increasing thus the comparability among groups and producing then reliable results. Randomisation will be done by a statistician not involved in the LEDEN study. Allocation concealment will be ensured by the use opaque sealed envelopes. The randomisation list will be kept in a electronic, password-locked file matching the name of participants to their research identification number and their group allocation. Given the nature of the LEDEN intervention (exercise training, i.e., a behavioural intervention) and since the main outcome measure, i.e., participants' ability in executing ADLs, of the the study requires that the outcome assessor knows how participants cope in their daily life, LEDEN is a unblinded study

Inclusion Criteria:

• Confirmed diagnosis of Alzheimer's disease, vascular or mixed dementia according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV);
• MMSE ≤ 20 (out of 30);
• living in one of the participating NHs for at least 3 months at the moment of baseline measurements (this time period will allow NH staff to have a good knowledge about the patient and about recent changes in his/her health status);
• to be able of walking at least 4 meters (with walking devices if needed but without human assistance);
• to be able to rise from a chair without help or with minimal human assistance.

Exclusion Criteria:

• Mild dementia, illustrated by a MMSE > 20 (out of 30)
• Terminal illness with life expectancy less than 6 months;
• Diagnosis of Parkinson's disease;
• Diagnosis of dementia with Lewy bodies;
• Unstable cardiovascular condition or any other health condition that might be deteriorated by physical exercise;
• Planned transfer from the NH to another NH/home or to surgery during the 6-month period of intervention;
• Already participating in physical exercise ≥ 2 times/week in the last 2 months prior to the date of baseline assessments.