摘要：Abstract Background To define a core set of geriatric data to be methodically collected in clinical cancer trials of older adults, enabling comparison across trials. Patients and methods Following a consensus approach, a panel of 14 geriatricians from oncology clinics identified seven domains of importance in geriatric assessment. Based on the international recommendations, geriatricians selected the mostly commonly used tools/items for geriatric assessment by domain (January–October 2015). The Geriatric Core Dataset (G-CODE) was progressively developed according to RAND appropriateness ratings and feedback during three successive Delphi rounds (July–September 2016). The face validity of the G-CODE was assessed with two large panels of health professionals (55 national and 42 international experts) involved both in clinical practice and cancer trials (March–September 2017). Results and discussion After the last Delphi round, the tools/items proposed for the G-CODE were the following: (1) social assessment: living alone or support requested to stay at home; (2) functional autonomy: Activities of Daily Living (ADL) questionnaire and short instrumental ADL questionnaire; (3) mobility: Timed Up and Go test; (4) nutrition: weight loss during the past 6 months and body mass index; (5) cognition: Mini-Cog test; (6) mood: mini-Geriatric Depression Scale and (7) comorbidity: updated Charlson Comorbidity Index. More than 70% of national experts (42 from 20 cities) and international experts (31 from 13 countries) participated. National and international surveys showed good acceptability of the G-CODE. Specific points discussed included age-year cut-off, threshold of each tool/item and information about social support, but no additional item was proposed. Conclusion We achieved formal consensus on a set of geriatric data to be collected in cancer trials of older patients. The dissemination and prospective use of the G-CODE is needed to assess its utility.