In 2020 the percentage of patients with a diagnosis of cancer in people with more than 65 years will exceed 70% and 28% in ethnic minorities. The treatment of cancer in these populations is challenging for the oncologists due to socio-economic issues such as poverty, reduced access to the hospital care, level of education. The clinical pathway "diagnosis-treatment-cure", typical of the care of young patients has to be integrated in elderly patients with a more individualized treatment by means of comprehensive geriatric assessment (CGA). IADL (Instrumental Activities of Daily Living) have the best predictive role in oncological setting and their impairment significantly correlate with overall survival, chemotherapy toxicities and thirty days postoperative morbidities. The CGA is universally accepted as the most appropriate instrument to analitically evaluate the age related problems of elderly patients. The role of CGA is crucial to identify geriatric issues not easily diagnosed, to predict treatment toxicities, functional or cognitive decline, post operative complications and to estimate life expectancy. The CGA items are predictive of severe toxicity, however it is not clearly established which are the best performers and the best cut-offs points. Today CGA is integrated with physical performance tests (the most widely used is the "time up and go" test) and laboratory assay of Interleukin 6 and D-Dimer that correlate with mortality and physical decline. There are few prospective studies that evaluated the role of CGA in treatment choice. The first is a phase II study in solid tumors, the second is a haematological trial on non Hodgkin lymphoma. The largest trial is a 571 patients observational series that confirmed the role of CGA in decision making. The administration of CGA is time consuming and consequently some screening tools were developed. VES-13 is a 13 items tool that explores prevalently the functional status and the self reported health status. VES-13 showed a good sensibility (87.3%) but a low specificity (62%) with respect to CGA for the diagnosis of patients with disabilities. Overcash et al. proposed an abbreviated form of CGA using a reduced number of items of ADL, IADL, MMSE and GDS. There was a good correlation between complete and reduced scales (coefficient of correlation 0.8). G8 is a screening tool composed of 8 questions that explore functional, cognitive and nutritional status. The score with the best equilibrium between sensibility and specificity was 14 (sensibility 85% and specificity 65%). In the first observational trial age, hystotype, chemotherapy dose, haemoglobin (man: 11 g/dL; women: 10 g/dL), creatinine clearance less than 34 mL/min (Jelliffe formula), earing problems, at least a fall in the last six months, walking problems, low social activity, were related to a major risk of toxicity; in another trial IADL, diastolic blood pressure, LDH and MAX2 index were predictive of haematological toxicity, while performance status, Mini-Mental Status score, Mini-Nutritional Assessment (MNA) score and MAX2 index were predictive of non haematological toxicity. Based on these parameters a 0-2 score was developed. A recent "position article" of EORTC (European organization for Research and Treatment of Cancer) and SIOG analyzed the pro and the contra of the use of some indicators in elderly patients. The overall survival (OS) frequently used in classical clinical trial could give wrong messages as there are some competitive risks of death in elderly patients. Another important indicator is the disease specific survival (DSS). Concerning the design of clinical trials, a possible strategy is to enrol elderly patients without upper age limit and to plan stratification. An interesting trial design is the so called "extended trial" that allow to re-open the arm of a trial in which a too low number of older patients was enrolled.