Enhancer mapping has been greatly facilitated by various genomic marks associated with it. However, little is available in our toolbox to link enhancers with their target promoters, hampering mechanistic understanding of enhancer-promoter (EP) interaction. We develop and characterize multiple genomic features for distinguishing true EP pairs from noninteracting pairs. We integrate these features into a probabilistic predictor for EP interactions. Multiple validation experiments demonstrate a significant improvement over state-of-the-art approaches. Systematic analyses of EP interactions across 12 cell types reveal several global features of EP interactions: (i) a larger fraction of EP interactions are cell type specific than enhancers; (ii) promoters controlled by multiple enhancers have higher tissue specificity, but the regulating enhancers are less conserved; (iii) cohesin plays a role in mediating tissue-specific EP interactions via chromatin looping in a CTCF-independent manner. Our approach presents a systematic and effective strategy to decipher the mechanisms underlying EP communication.T ranscriptional enhancers represent the primary basis for differential gene expression. These elements regulate cell type specificity, development, and metazoan evolution, with many human diseases resulting from altered enhancer action (1, 2).A key gap in our knowledge is an understanding of how enhancers select specific promoters for activation. Linkage of enhancers and target promoters is challenged by enhancer properties. First, increasing evidence suggests that enhancers are not located adjacent to their target promoters. Instead, they are positioned tens of kilobases away and contact their targets via long-range interactions (3-6). Second, enhancers are position independent, i.e., they may be located either upstream or downstream of the regulated promoter.Experimental approaches to identifying enhancer targets have largely relied on chromosome conformation capture (3C) (7) and its variants such as circularized chromosome conformation capture (4C) and genome-wide chromosome conformation capture (Hi-C) (8), all of which determine the relative frequency of direct physical contact between linearly separated DNA sequences. Unlike 3C and 4C, Hi-C is a truly genome-wide technology, but its current resolution (1 Mbp) in general is not high enough to distinguish individual enhancer-promoter (EP) interactions (9). Newer methods such as ChIP-loop (10) and chromatin interaction analysis with paired-end tag sequencing (ChIA-PET) (11) combine the principles of 3C and ChIP to identify chromatin interactions mediated by protein factors. However, the assays are technically challenging and currently have a high false-negative rate (5, 12). Therefore, computational work, if successful, can complement experimental protocols and allow prioritization of future experiments much more effectively.The most common computational approach is assigning the nearest promoter of an enhancer as its target. Improvements to this basic approach have be...