Altered RNA splicing is implicated in many human diseases, although how common this phenomenon is and to what extent genetic variation contributes to it is not known. To address these questions, Brendan Frey and colleagues developed a machine learning algorithm that predicts the impact of single-nucleotide variants (SNVs) on RNA splicing and analyzed the association of variants with disease (Science doi:10.1126/science.1254806; 18 December 2014). Using this approach, they detected thousands of variants that might be involved in a myriad of diseases, including spinal muscular atrophy, colorectal cancer and autism spectrum disorder (ASD). Indeed, a genome-wide analysis of several individuals with ASD showed that, for the genes predicted to be misspliced, there was enrichment for transcripts that are highly expressed in brain tissue and that are associated with neurodevelopmental conditions. They also discovered that intronic mutations linked to disease are nine times more likely to affect splicing than common ones and that missense exonic mutations with little impact on protein function are five times more likely to perturb normal splicing. It will be interesting in the future to incorporate additional layers of splicing regulation into computational models, such as epigenetic modifications and chromatin remodeling.