Genomic systematic evolution of ligands by exponential enrichment (Genomic SELEX) is an experimental procedure for the expression condition-independent identification of protein-binding RNAs. RNA libraries derived from genomic DNA are generated via random priming, PCR amplification and in vitro transcription. Libraries consist of genomic sequences of selected size, and fragments are flanked by constant sequences required for amplification and transcription. This RNA pool is then subjected to several rounds of selection and amplification to enrich for RNAs meeting the selection criteria. Various selection criteria are possible. Here we describe selection by affinity to a protein of interest. High-affinity ligands can then be cloned and sequenced to allow their identification. With this method, protein-binding RNAs can be discovered, nucleic acid–protein interactions can be identified, and whole protein–nucleic acid networks can be defined. This method is also suitable for discovering novel genes, including non-protein-coding RNAs, and it complements in silico approaches. It is better suited to detect protein-binding RNAs that are differentially expressed (and therefore absent from many tissues) and low-abundance RNAs than experimental procedures that start from the isolation of expressed RNAs. The protocol takes ∼3 months to complete.