Hepatitis C virus (HCV) nonstructural protein 5A (NS5A) exhibits a preference for G/U-rich RNA in vitro. Biological analysis of the NS5A RNA-binding activity and its target sites in the genome will be facilitated by a description of the NS5A-RNA complex. We demonstrate that the C-4 carbonyl of the uracil base and, by inference, the C-6 carbonyl of the guanine base interact with NS5A. U-rich RNA of 5 to 6 nucleotides (nt) is sufficient for high-affinity binding to NS5A. The minimal RNA-binding domain of NS5A consists of residues 2005 to 2221 (referred to as domain I-plus). This region of the protein includes the amino-terminal domain I as well as the subsequent linker that separates domains I and II. This linker region is the site of adaptive mutations. U-rich RNA-binding activity is not observed for an NS5A derivative containing only residues 2194 to 2419 (domains II and III). Mass spectrometric analysis of an NS5A-poly(rU) complex identified domains I and II as sites for interaction with RNA. Dimerization of NS5A was demonstrated by glutaraldehyde cross-linking. This dimerization is likely mediated by domain I-plus, as dimers of this protein are trapped by cross-linking. Dimers of the domain II-III protein are not observed. The monomer-dimer equilibrium of NS5A shifts in favor of dimer when U-rich RNA is present but not when A-rich RNA is present, consistent with an NS5A dimer being the RNA-binding-competent form of the protein. These data provide a molecular perspective of the NS5A-RNA complex and suggest possible mechanisms for regulation of HCV and cellular gene expression.