BACKGROUND: Chronic obstructive pulmonary disease (COPD) is a multi-system disorder driven partly by diffuse inflammation, malnutrition and haematological aberrations. Because the red blood cell distribution width (RDW) is a surrogate of these anomalies, we hypothesised that it might be of prognostic importance in COPD patients. Additionally, we tested the supposition that iron deficiency (ID) per se could be a prevalent and ominous co-morbidity in these individuals. METHODS: We analysed the relation of red cell indices on admission and over time with mortality in 655 consecutively eligible COPD patients (mean±SD age 77±12 y, FEV1 0.85±0.34 l, FVC 1.44±0.60 l, 54% male). Concomitant heart failure, ischaemic heart disease, and neoplasia were exclusion criteria. The combination of a high RDW and low mean cell haemoglobin (MCH) was utilised to identify ID. RESULTS: On admission, an RDW>15%, Hb<12.5 g/dl, MCH<27, and ID were evident in 33%, 31%, 12% and 10% of patients. Compared to those with an RDW≤15%, patients with levels >15% had lower Hbs, lower FEV1s, and longer median (±IQR) hospital stays (9±11 vs 8±8 days, P<0.001). Over a mean period of 40±29 months, 227 (35%) patients died. On Cox proportional hazards analyses, a higher RDW predicted increased mortality (adjusted χ 16, P<0.0001) independently of age (χ 11, P<0.001), FEV1 (χ 5, P<0.03), Hb and creatinine (latter two not retained in model) and provided graded prognostic information () incremental to that of FEV1 (P<0.05 for change in χ). Over time, 63%, 72%, 65%, and 46% of patients had a rise in RDW, a fall in Hb, a fall in MCH, and evolving ID (rising RDW and falling MCH), respectively. A rising RDW predicted death (adjusted χ 32, P<0.0001) independently of baseline RDWs and changes in Hb, with an increase greater than 0.03% per month associated with a twofold escalated risk of mortality (). Evolving ID was also associated with poorer survival ().(Figure is included in full-text article.) CONCLUSIONS: An elevated RDW alone and iron deficiency predict an amplified risk of death in COPD and could be utilised for risk stratification or therapeutically targeted to improve outcomes.