Background: TRIM5α is a cytosolic pattern-recognition receptor that intercepts incoming capsids of retroviruses, including HIV-1. Capsid recognition results in species-specific restriction of viral replication and induction of an innate immune response. The prevailing model is that TRIM5α recognizes the HIV capsid through synergy of at least 3 biochemical activities: direct binding of capsid subunits by a B30.2/SPRY domain, TRIM5α dimerization, and higher-order assembly of the tripartite motif into a hexagonal lattice. A key point is that multiple B30.2 domains in the TRIM5α assembly are positioned to match the spacing of their repeating binding epitopes on the capsid surface, thereby vastly amplifying the avidity of capsid recognition. However, the molecular details of the above biochemical activities remain to be uncovered. The tripartite motif is a series of domains – RING, B-box, and coiled-coil – arranged in that order from the N-terminus. Here, we describe the structure of the coiled-coil, which mediates TRIM dimerization and is required for higherorder TRIM5α assembly. Our studies reveal how TRIM5α dimerizes, how it positions the B30.2/SPRY domains to bind capsid subunits, and how TRIM5α dimers organize into hexagonal arrays to recognize the capsid lattice and restrict viral replication. Methodology: Working under the framework that a comprehensive mechanistic model must address how the various TRIM domains are structurally coupled, we initially focused on the coiled-coil, because it specifies dimerization and defines the spatial relationships between the other domains. We used human TRIM25 as a model system since it lacks TRIM5’s propensity for higher-order assembly and is therefore more tractable biochemically. We determined the crystal structure of the TRIM25 coiled-coil at 2.6 A resolution. Based on this structure, we built and experimentally validated a homology model for the TRIM5α coiled-coil. Results: Our analyses revealed that: 1) the TRIM coiled-coil is an elongated dimer of hairpin-shaped subunits; 2) the coiled-coil dimers form the edges of the TRIM5α hexagon, positioning two C-terminal B30.2 domains at the center of each edge and the N-terminal RING and B-box domains at the vertices; and 3) the “L2” linker between the coiled-coil and B30.2 is integrated with the tripartite motif and participates in dimer formation. We propose that the L2 linker of TRIM5α structurally couples the tripartite motif to its B30.2 domain, thereby orienting the two capsid-binding domains of one TRIM5α dimer on the same side and positioning them with a defined spacing relative to each other. Conclusions: Our studies support a model wherein TRIM5 proteins recognize retroviral capsids by acting as “molecular rulers” that measure the spacing of the assembled capsid subunits.