We are grateful to Dr Martin et al for their comments, and accept that our study had certain design limitations. The data were obtained during the run-in for another study (Cowan et al, Thorax Published Online First: 23 September 2010. doi:10.1136/thx.2010.144592). However, the principal finding remains: while we agree that the presence of airway eosinophilia is a reliable predictor of steroid responsiveness, the absence of eosinophilia does not accurately predict steroid unresponsiveness. Whether intentionally or not, these authors imply that only patients with demonstrable sputum eosinophilia are steroid responsive. This is not the case.