Objective: To investigate the factor structure of a ten-item version of the Center for Epidemiologic Studies Depression Scale (CESD-10). Method: Two hundred and thirty-one Chinese persons aged 60–92 were individually interviewed with the CESD-10. The data were subject to confirmatory factor analysis testing several competing models. The one-factor model hypothesizes that all ten items are loaded on a single factor. In the two-factor model, positive affect items are loaded on one factor, and the remaining items on the other. The three-factor model distinguishes between positive affect, depressed affect, and somatic symptoms. Results: The three-factor model was the best fitting model, with a comparative fit index of 0.95 and a standardized root mean square residual of 0.06. Depressed affect and somatic symptoms were highly correlated, and both were moderately correlated with positive affect. Nonetheless, all three factors appeared to tap a common underlying construct of depression—when a higher-order construct of depression was allowed to explain the intercorrelations of the three factors, depressed affect loaded at 0.88, somatic symptoms loaded at 0.92, and positive affect loaded at 0.51, on the higher-order construct. Conclusion: The data provided support for the factorial validity of the CESD-10.