Currently, with exception of lung function tests, there are no well validated biomarkers or surrogate endpoints that can be used to establish efficacy of novel drugs for chronic obstructive pulmonary disease (COPD). However, the lung function test is not an ideal surrogate for short-term drug trials because it (1) does not provide information regarding disease activity or the underlying pathologic process, (2) cannot separate the various phenotypes of COPD, (3) is not specific for COPD, and (4) is relatively unresponsive to known therapies that prolong survival. Accordingly, there are large-scale studies presently underway to identify novel biomarkers in COPD. In this article, we discuss the current barriers of biomarker discovery and propose possible criteria and methods for developing novel biomarkers in COPD.