Number of elderly people over 70 is increasing in France. In 2002, in the south-west, 273,741 of them were over 75 years, i.e. 9.2% of Aquitaine population. Incidence of cancer in this old age is 16,500 new cases per year, more than fifty percent of people with cancer. Treatment modalities of elderly people is one of the 70 priorities of the "Plan cancer" instituted by the French government. They are very few therapeutic trials dedicated. Oncologists hesitate to treat them because they are either afraid of inducing toxicity or of breaking down quality of life. Consequently, we decided to launch a protocol with both oncologists and geriatricians which principal aim is to find out if geriatric assessment data can help to better predict for chemotherapy toxicity, loss of autonomy and survival. We plan to accrue 360 patients diagnosed for cancer, including digestive, pulmonary, prostate, lymphoma, bladder, ovary cancer for whom first-line chemotherapy is planned. Patients are initially classified according to usual methods of medical oncology practice into three groups: patients who can receive standard treatment, reduced standard treatment or treatment adapted to the frail condition. Around Aquitaine, , we organised seven teams composed of one geriatrician and one nurse. Two kind of teams were activated: one which cover ten treatment sites in Bordeaux area and six sedentary teams which worked half a day a week in designated hospitals . Geriatric evaluation included test of cognitive functions (MMS), nutritional status (MNA), co-morbidity (CIRS-G), mobility (Get up and Go), activities (ADL;IADL), quality of life (QLQ-C30), depression (GDS-15) and Lachs-Balducci screening. Patients have four geriatric evaluations : before treatment, day 1 cycle 2, day 1 cycle 4, day 1 cycle 7 and/or end of chemotherapy. Since September 2002, 177 patients have been included, 112 have finished: 47.3% have received four evaluations, 16.1% died before the end of protocol, 14.3% stopped because they were in progression and changed their treatment, 11.6% met administrative problem that didn't allow all evaluations, 7.1% declined after inclusion and 3.6% finished their treatment before. The following results have been obtained: before treatment, 73% of these patients were at risk of undernutrition (MNA< 23.5), about 1/3 had one or more inability or a risk of falls (38% IADL<6, 29% get up and go>20seconds, 27% ADL>1, 34% PS<1), 28% of them had altered cognitive functions (MMS<24), 29% were depressive (GDS-15>6), 25% thought they had poor quality of life (QLQ-C30<4). Protocol will be closed in September 2005.