Aim The Charlson Comorbidity Index (CCI) is commonly studied for predicting mortality, but there is no validation study of it in Chinese older adults. The objective of the present study was to validate the use of CCI in Chinese older adults for predicting mortality. Method We carried out a retrospective cohort study from 2004 to 2013 for patients discharged from a geriatric day hospital in Hong Kong. Comorbidity was quantified using CCI, and patients were divided into six groups according to their score of CCI: CCI‐0, CCI‐1, CCI‐2, CCI‐3, CCI‐4 and CCI ≥ 5. Other data collected included demographics, and functional, nutritional, cognitive and social assessment. The outcome measure was 1‐year mortality. Results At 1‐year follow up, 3.8% (n = 17), 5.9% (n = 37), 9.2% (n = 35), 12.9% (n = 20), 16.9% (n = 23) and 19.3% (n = 60) of CCI‐0, CCI‐1, CCI‐2, CCI‐3, CCI‐4 and CCI ≥ 5 died, respectively (P < 0.001). Multivariate analysis showed that CCI‐1, CCI‐2, CCI‐3, CCI‐4 and CCI ≥ 5 have a hazard ratio (HR) of 1.34 (confidence interval [CI] 1.04–2.12), 2.18 (CI 1.03–4.61), 3.44 (CI 1.52–7.81), 3.74 (CI 1.35–10.39) and 4.63 (CI 2.28–9.43), respectively, compared with CCI‐0. The area under the curve of the receiver operating characteristic curves of CCI in predicting 1‐year mortality for older adults was 0.68 (CI 0.64–0.72). Conclusion There is a significant dose–response relationship in the hazard ratio between CCI and 1‐year mortality in Chinese older adults, but involvements of functional, nutritional and social assessments are important for comprehensive quantification of health status in older adults. Geriatr Gerontol Int 2014; 14: 452–457.