Objectives The cutoff scores for the Geriatric Depression Scale (GDS) commonly adopted in clinical and research settings are based upon other neuropsychological tests. However, any intervention for depression should aim at improving subjective quality of life (QoL). We searched for a GDS cutoff level that might identify a decrease in perceived QoL using a scale that also allows formal cost‐effectiveness calculations. Methods Quality of life was assessed by the Health Utilities Index, Mark 3 in all 344 residents of Tuscania (Italy) aged 75 years and above. Mood was assessed by both the 30‐item GDS and the derived 15‐item GDS. The association of GDS with low QoL was analyzed by multivariable logistic regression. Receiver operating characteristic curve analysis was adopted to estimate the overall predictive value and the best GDS cutoff for poor QoL. Results The 30‐item GDS score was associated with increased probability of a worse QoL (odds ratio (OR) = 1.07, 95% confidence (CI) = 1.02–1.12, p = 0.003); also, it was a fair predictor of worse QoL (area under the curve (AUC) = 0.72; 95% CI = 0.67–0.76). The best GDS score cutoff for identifying a poor QoL was above 9/30. Results were similar (OR = 1.07, 95% CI = 1.02–1.12, p = 0.003, and AUC = 0.72, 95% CI = 0.67–0.76) for the short GDS form for a cutoff above 5/15. Conclusions Among older subjects, depressive symptoms are associated with reduced QoL; GDS scores above 9/30 or 5/15 best predict poor perceived health‐related QoL. These cutoff scores could therefore identify subjects in whom treatment is more likely to improve QoL and to yield a favorable cost‐effectiveness ratio. Copyright © 2017 John Wiley & Sons, Ltd.