Continental rifts are long narrow features bounded on one or both sides by normal faults along which several kilometers of displacement has occurred, resulting in deep rift valleys that form sedimentary basins. Continental rifts all share the same geometrical features: They have widths that are characteristic of the region in which they form, they are predominantly asymmetric, and they are segmented along axes at a length scale that is also characteristic of the region. We describe a very simple mechanical model to explain these three features. The rift initiates with the formation of a conjugate set of normal faults that intersect at the center of the seismogenic layer. One of these becomes the boundary fault system; the other, the failed conjugate fault (being offset by the former), becomes locked early in the rifting history. Hence an asymmetric graben is formed, which is the basic building block of rifts. The failed conjugate fault has distinctive properties that allow it to be easily distinguished from other faults. Examples of these faults from the Tanganyika and Malawi rifts show excellent agreement with predictions about their positions with respect to the boundary fault system and their net displacement and slip history. The width of the asymmetric graben is a simple function of seismogenic thickness. The net displacement on the boundary fault system is limited by the resistive friction and the flexural restoring force. We calculate this limiting boundary fault system displacement and, using the displacement-length scaling law for normal faults, suggest that this controls the along-axis segmentation length. Both maximum fault slip and segmentation length are functions of the seismogenic thickness and the effective elastic thickness of the crust. These calculations agree very well with observations of segmentation from the African rifts and the maximum length of range-fronting normal faults in Nevada.