Lung cancer is the leading cause of cancer-related deaths worldwide. High-throughput technologies such as microarrays provide an opportunity to explore biomarkers for cancer prevention, prognosis and treatment guidance. Recent studies have revealed many biomarkers with the potential for clinical application. However, major limitations still exist. Although useful data on cancer genomics has accumulated rapidly, there has also been a simultaneous tendency for amplification of the complex relationships among the enormous number of variables that need to be considered. Disentangling these complex gene–gene interactions requires new approaches to data analysis to reveal information that has been obscured by traditional methods. Here, we review the current findings on biomarker identification in lung cancer, address their limitations and discuss some future directions for improvements in this area of research.