N6-methyladenosine (m6A) is the most abundant RNA modification in eukaryotes, present in a large subset of the transcriptome. m6A is thought to act as a dynamic switch that affects processes such as RNA degradation, localization and splicing. Although 'writers', 'erasers' and 'readers' of this mark have been identified, the mode by which m6A exerts its functions is still poorly understood. Parisien, Pan and colleagues now demonstrate that m6A controls RNA-protein interactions by altering local RNA structure. The authors initially identify heterogeneous nuclear ribonucleoprotein C1/C2 (hnRNP C) as a protein that preferentially interacts with an m6A-modified hairpin originating from the long noncoding RNA MALAT1. hnRNP C affects pre-mRNA processing and binds nascent RNA transcripts via single-stranded U tracts. Curiously, the hnRNP C–binding site in the MALAT1 hairpin is located opposite the m6A-modification site. Adenosine methylation is known to destabilize RNA duplexes, and indeed m6A modification of the MALAT1 hairpin increased the accessibility of its opposing U tract. Transcriptome-wide analyses indicated that these 'm6A switches' are common: 7% of all hnRNP C–binding sites corresponded to an m6A-modified consensus motif located in the vicinity of a poly(U) tract. hnRNP C PAR-CLIP experiments in cells expressing siRNAs targeting METTL3 and METTL14, two enzymes involved in m6A deposition, demonstrated that a substantial number of the identified m6A switches showed decreased hnRNP C binding upon m6A depletion. Subsequent analysis of these 'high-confidence' m6A switches supported the finding that m6A influences hnRNP C binding by altering RNA structure, with striking effects on RNA abundance and splicing. The identification of m6A as a structural remodeler opens a new perspective on its mechanism of action, and it will be interesting to learn how many other RNA-binding proteins are affected by this mechanism. (Nature doi:10.1038/nature14234, 26 February 2015).