Although long noncoding RNAs (lncRNAs) have been recognized in recent years to constitute a significant portion of mammalian transcriptome, and the functional impact of several lncRNAs has been characterized by a few studies, yet it is still difficult to large-scale ascertain their functions from lab experiment or structure prediction. To address this deficit, we have developed a computational pipeline to large-scale annotate the functions of lncRNA based on coding-noncoding gene co-expression network. In this network, a node (circle) represents the protein-coding gene or lncRNA, and an edge (connecting line between nodes) indicates that the two genes are co-expressed (the correlation coefficients of connected genes reached the defined cutoff). In this chapter, we show how to use an lncRNA functional annotation pipeline to construct a co-expression network based on gene expression profiles in prostate cancer and how to further predict lncRNA functions using model-based and hub-based sub-networks.