Hospital admissions following emergency visits of elderly people are frequent. This admission modality is often problematic both for the patients and the emergency healthcare professionals. Direct admission from home (or nursing home) in acute geriatric units (AGU) has been developed but has never been prospectively assessed. We conducted a 6-month prospective observational study to compare the 97 patients admitted through the emergency room (ER) in the AGU of Bichat's hospital to the 76 patients admitted directly. Collected data included socio-demographic and medical baseline data, clinical severity score at admission, cause of hospitalization, final diagnosis, in-hospital occurrence of urinary retention and of pressure ulcer, length of stay, discharge disposition and mortality. No significant differences between the groups were found for most baseline characteristics, clinical severity score, occurrence of pressure ulcers, length of stay and mortality. However ER patients were significantly older (88±6 vs 86±7 years, p=0.04) and had more often history of arrhythmia (29% vs 15%, p=0.02) and protein-energy malnutrition preceding admission (63% vs 46%, p=0.03). Falls as admission cause was more common in ER patients while unexplained health status or functional decline were most common in those admitted directly. Clinical outcomes were less favourable in ER patients with significantly more urinary retentions (25% vs 4%, p=0.0002) and transfers to rehabilitation units (48% vs 31%, p=0.04). The patients admitted directly returned more often at home without additional social support (53% vs 30%, p=0.001). Direct admission in AGU is feasible, medically effective and provides an alternative to attending an emergency room. This admission modality could be specially suitable for elder people suffering from an unexplained functional or health status declines. Further studies are necessary to support the hypothesis that quality gains and cost-effective measures may be achieved by dissemination of such an admission modality at the hospital.