Background: We sought to determine whether circulating inflammatory biomarkers were associated with the frequent exacerbator phenotype in stable COPD patients ie, those with two or more exacerbations in the previous year. Methods: Eighty-eight stable, severe, COPD patients (4 females) were assessed for exacerbation frequency, pulmonary function, fraction of expired nitric oxide (F E NO); inflammatory variables were measured in venous blood. Logistic regression assessed associations between the frequent exacerbator phenotype and systemic inflammation. Results: Compared with infrequent exacerbators, frequent exacerbators (n=10; 11.4%) had greater serum concentration (median (25th-75th quartile)) of serum amyloid A (SAA; 134 (84-178) vs 71 (38-116) ng/mL; P=0.024), surfactant protein D (SP-D; 15.6 (9.0-19.3) vs 8.5 (3.6-14.9) ng/mL; P=0.049) and interleukin-4 (IL-4; 0.12 (0.08-1.44) vs 0.03 (0.01-0.10) pg/mL; P=0.001). SAA, SP-D and IL-4 were not significantly correlated with FEV 1 % predicted or FVC %predicted. After adjusting for sex, age, BMI, FEV 1 /FVC and smoking pack-years, only SAA remained independently associated with the frequent exacerbator phenotype (OR 1.49[1.09-2.04]; P=0.012). The odds of being a frequent exacerbator was 18times greater in the highest SAA quartile (≥124.1 ng/mL) than the lowest SAA quartile (≤44.1 ng/mL) (OR 18.34[1.30-258.81]; P=0.031), and there was a significant positive trend of increasing OR with increasing SAA quartile (P=0.008). For SAA, the area under the receiver operating characteristic curve was 0.721 for identification of frequent exacerbators; an SAA cutoff of 87.0 ng/mL yielded an 80% sensitivity and 61.5% specificity. Conclusion: In stable COPD patients, SAA was independently associated with the frequent exacerbator phenotype, suggesting that SAA may be a useful serum biomarker to inform progression or management in COPD.