Cytoskeletal filaments are often capped at one end, regulating assembly and cellular location. The actin filament is a right-handed, two-strand long-pitch helix. The ends of the two protofilaments are staggered in relation to each other, suggesting that capping could result from one protein binding simultaneously to the ends of both protofilaments. Capping protein (CP), a ubiquitous alpha/beta heterodimer in eukaryotes, tightly caps (K(d) approximately 0.1-1 nM) the barbed end of the actin filament (the end favored for polymerization), preventing actin subunit addition and loss. CP is critical for actin assembly and actin-based motility in vivo and is an essential component of the dendritic nucleation model for actin polymerization at the leading edge of cells. However, the mechanism by which CP caps actin filaments is not well understood. The X-ray crystal structure of CP has inspired a model where the C termini ( approximately 30 amino acids) of the alpha and beta subunits of CP are mobile extensions ("tentacles"), and these regions are responsible for high-affinity binding to, and functional capping of, the barbed end. We tested the tentacle model in vitro with recombinant mutant CPs. Loss of both tentacles causes a complete loss of capping activity. The alpha tentacle contributes more to capping affinity and kinetics; its removal reduces capping affinity by 5000-fold and the on-rate of capping by 20-fold. In contrast, removal of the beta tentacle reduced the affinity by only 300-fold and did not affect the on-rate. These two regions are not close to each other in the three-dimensional structure, suggesting CP uses two independent actin binding tentacles to cap the barbed end. CP with either tentacle alone can cap, as can the isolated beta tentacle alone, suggesting that the individual tentacles interact with more than one actin subunit at a subunit interface at the barbed end.