Broadcasting is used in on-demand routing protocols to discover routes in Mobile Ad-hoc Networks (MANETs). On-demand routing protocols, such as AODV (Ad-hoc On-demand Distance Vector) routing, commonly employ pure flooding based broadcasting. However, pure flooding generates excessively redundant routing traffic that may lead to broadcast storm problem (BSP) and deteriorate the performance of MANETs significantly. Probabilistic broadcasting schemes were proposed in the literature to address BSP. However, these schemes do not consider thermal noise and interference which exist in real life MANETs, and therefore, do not perform well in real life MANETs. This paper presents a novel Channel Adaptive Probabilistic Broadcast (CAPB) scheme to disseminate RREQ packets efficiently. The proposed CAPB scheme determines the probability of rebroadcasting RREQ packets on the fly according to the current SINR (Signal to Interference plus Noise Ratio) and node density in the neighborhood. The proposed scheme and two related state of the art (SoA) schemes from the literature (  and ) are implemented in the standard AODV routing protocol to replace the pure flooding based broadcast. Simulation results show that the proposed scheme outperforms the standard AODV and the two competitors in terms of routing overhead, throughput and end-to-end delay significantly.