The reaction of (Cp*ReH₂)₂B₄H₄ with monoborane leads to the sequential formation of (Cp*Re)₂B_nH_n (n = 7−10, 1−4). These species adopt closed deltahedra with the same total connectivities as the closo-borane anions [B_nH_n]^2-, n = 9−12, but with flattened geometries rather than spherical shapes. These rhenaborane clusters are characterized by high metal coordination numbers, Re−Re cross-cluster distances within the Re−Re single bond range, and formal cluster electron counts three skeletal electron pairs short of that required for a canonical closo-structure of the same nuclearity. An open cluster, (Cp*ReH)₂B₇H₉ (5), is isolated that bears the same structural relationship to arachno-B₉H₁₅ as 1−4 bear to the closo-borane anions. Chloroborane permits the isolation of (Cp*ReH)₂B₅Cl₅ (6), an isoelectronic chloro-analogue of known open (Cp*WH₂)₂B₅H₅ and (Cp*Re)₂B₆H₄Cl₂ (7), a triple-decker complex containing a planar, six-membered 1,2-B₆H₄Cl₂ ring. Both are putative five- and six-boron intermediates in the formation of 1. Electronic structure calculations (extended Hückel and density functional theory) yield geometries in agreement with the structure determinations, large HOMO−LUMO gaps in accord with the high stabilities, and ¹¹B chemical shifts accurately reflecting the observed shifts. Analyses of the bonding in 1−4 reveal that the Cp*Re···Cp*Re interaction generates fragment orbitals that are able to contribute the “missing” three skeletal electron pairs required for skeletal bonding. The necessity of a Re···Re interaction for strong cluster bonding requires a borane fragment shape change to accommodate it, thereby explaining the noncanonical geometries. Application of the debor principle of borane chemistry to the shapes of 1−4 readily rationalizes the observed geometries of 5 and 6. This evidence of the scope of transition metal fragment control of borane geometry suggests the existence of a large class of metallaboranes with structures not found in known borane or metal clusters.