The formation of the Pamir salient is related to pre-existing arcuate weak zones. However, the mechanism for the reactivation of the weak zones during the early stage of the Indo-Eurasian collision remains unknown. In this study, we apply analogue modeling to address this issue. Based on similarity principle of scales, we have set up two models with pre-existing weak zone in combination of either non-décollement or décollement. Our modeling yields the following outcomes. ① The modeling result validates the conception that a pre-existing arcuate weak zone can produce the Pamir salient. ② Under the condition of a pre-existing arcuate weak zone without décollement, deformation propagates northward, with the strain loaded in the pre-existing weak zone during late deformation stage. Southdipping thrust faults propagate forward, widening the fold-and-thrust belt. The topography exhibits a slope dipping to the north, showing a northward decreasing wedge. ③ Under the condition of a pre-existing arcuate weak zone and décollement, the former was reactivated at a very early deformation stage, with the strain immediately loaded in the weak zone. Subsequently, the deformation progressively propagates northward from the moveable wall. When the shortening amount is large enough, frontal thrust emerges at front of the weak zone. The deformation architecture consists of several popup structures, resulting in a topography characterized by alternatively arranged ranges and valleys. ④ A décollement is necessary for the strain be exerted immediately on the arcuate weak zone at the northern boundary of the Pamir in the initial stage of the Indo-Eurasian collision. The confirmation of the existence, depth and interface of the early Cenozoic décollement requires further studies.