OBJECTIVES Chronic Obstructive Pulmonary Disease (COPD) is a progressive condition which is characterized by a dramatic socio-economic impact. Sensitivity of clinical signs and lung function in predicting death is variable in different COPD phenotypes. Aim: to assess the predictive value of COPD annual cost on mortality.  METHODS : gender; age; smoking habit, clinical data, and complete lung funtion were assessed in 275 consecutive COPD patients aged >40y together with the annual cost calculated over the last twelve months. Statistics: t test for comparing means ± sd; linear regression for checking any relationship between each variable and mortality (p<0.05 was accepted).  RESULTS: The whole mortality was 40.4% over three years (n=12; 47, and 52 subjects, respectively). Subjects still survived after three years (n=164) proved originally different from those (n=111) who died  in terms of mean age; FEV1 (in l); RV; TLCO/VA; 6’ walking test; BODE index, and Charlson index (all p<0.001), but not of FEV1% pred. and FEV1/FVC (p=ns). Mean total COPD cost was [95%CI: [95% CI: 968.4; 1,353.6] in survivors and [95% CI: 5,508.0; 6,809.8] in those who died, respectively (p<0.001). The hospitalization cost impacted for 78.2% of the total annual cost in subjects who died, the absolute value being sixfold higher than in survivors. All economic components of cost were discriminant at three years, independently of gender.  CONCLUSIONS : 1) several clinical and lung function variables contribute to predict mortality in COPD; 2) identification of COPD phenotypes is crucial, but multiple indices are required; 3) total annual cost proved the most sensitive predictor of mortality at three years; 4) annual cost is much easier and cheaper to obtain over twelve months; 5) data are supporting the high predicting value and the convenience of COPD “economic phenotyping”.