Background: Periostin levels are associated with airway eosinophilia and are suppressed by corticosteroid treatment in asthma. This study sought to determine the relationship between serum and sputum periostin, airway inflammatory phenotype and asthma control.Methods: Adults with poorly-controlled asthma (n = 83) underwent a clinical assessment, sputum induction and blood sampling. Dispersed sputum was used for a differential cell count and periostin assessment (ELISA). Serum periostin was determined by the Elecsys (R) immunoassay.Results: Periostin levels were significantly higher in serum (median (IQR) of 51.6 (41.8, 62.6) ng/mL) than in sputum (1.1 (0.5, 2.0) ng/mL) (p <0.001). Serum and sputum periostin were significantly higher in patients with eosinophilic asthma (n = 37) compared with non-eosinophilic asthma. Both serum and sputum periostin levels were significantly associated with proportion of sputum eosinophils (r = 0.422, p <0.001 and r = 0.364, p = 0.005 respectively) but were not associated with asthma control. In receiver operator characteristic curve analysis, the area under the curve (AUC) for serum periostin (n = 83) was 0.679, p = 0.007. Peripheral blood eosinophils assessed in 67 matched samples, had a numerically greater AUC of 0.820 compared with serum periostin, p = 0.086 for the detection of eosinophilic asthma.Conclusion: In poorly-controlled asthma, sputum and serum periostin levels are significantly related to sputum eosinophil proportions while their ability to predict the presence of eosinophilic asthma is modest.