Cenozoic rifts in Tibet were traditionally interpreted as a result of topographic collapse of the Tibetan Plateau, reaching the maximum elevation that can be supported by its mechanical strength. Recent studies have emphasized possible similarities between rifting in Tibet and extension in the Basin and Range of the western United States. However, when examined in detail, one finds that spacing of long (>100 km) rifts in Tibet (~100--300 km) is significantly greater than that in the Basin and Range (~20--40 km). From south to north, rift spacing decreases systematically: from 191¿67 km in the Himalaya south of the Indus-Yalu suture to 146¿34 km in south Tibet between the Indus-Yalu and Bangong-Nujiang sutures and farther north to 101¿31 km in central Tibet between the Bangong-Nujiang and Jinsha sutures. Instability analysis suggests that the mantle lithosphere must have been involved in east-west Tibetan extension. Specifically, the widely spaced rifts in The Himalaya and Tibet may have been related to the presence of a relatively light crust (density <~2.90 g cm-3) and a strong mantle lithosphere (~40 km thick and a factor of 5 stronger than the upper crust). The observed systematic decrease in rift spacing can be explained by the known decrease in the crustal thickness in Tibet, from ~70--80 km in The Himalaya in the south to ~50--55 km in central Tibet in the north. A regional comparison of rifts in east Asia suggests that both the involvement of the mantle lithosphere and the age of rift initiation are similar for Tibetan rifts, the Baikal rift, and the Shanxi graben. This implies that topographic collapse or a convective event in the mantle cannot be the sole cause for the development of the Tibetan rifts. A regional boundary condition applied throughout east Asia must be required. ¿ 2000 American Geophysical Union