This study evaluated the efficacy of two different group procedures of non-pharmacological treatment in mild-to-moderate Alzheimer's disease (AD). Thirty-two patients entered the study and were divided in groups of four subjects. We compared recreational activities ('global' stimulation) with a combination of procedural memory training on activities of daily living and neuropsychological rehabilitation of 'residual' functions ('cognitive-specific'). All patients and caregivers were ensured psychological support. Both group treatments were delivered for six weeks. Multidimensional efficacy assessment of functional, behavioural and neuropsychological aspects was performed. Patients receiving 'global' stimulation showed a substantial reduction in behavioural disturbances (Neuropsychiatric Inventory [NPI]: frequency p = 0.034; severity p = 0.012); Revised Memory Behaviour Problems Checklist (frequency p = 0.008; reaction p = 0.027), and better performance in the Functional Living Skills Assessment (FLSA), a standardized direct measure of performance in everyday life (p = 0.021) and Verbal Fluency for Letters (p = 0.000). Patients receiving 'cognitive-specific' treatment improved only on the scale evaluating functional competence in daily living (Nurses' Observation Scale for Geriatric Patients [NOSGER] p = 0.018). At follow-up (six months later), compared with baseline, patients following the 'global' stimulation treatment showed an improvement at caregiver distress on NPI (p = 0.04). No other significant difference was detected. Our results support the contention that a 'global' treatment can lead to a significant improvement in AD patients, both for behavioural and functional aspects. The 'cognitive-specific' treatment we used in this research did not show better efficacy.