The TCP – Transmission Control Protocol provides connection oriented and reliable process to process communication. TCP also performs flow control, error control and congestion control. TCP’s default interpretation of any loss as a side effect of network congestion works well with the wired networks. But when Standard TCP is used with the wireless networks, especially with the MANETs Mobile Adhoc Networks, it becomes a reason of performance degradation. MANETs may suffer from channel losses, route failure losses and congestion losses. Unfortunately it is not possible to differentiate these losses without predictions. Highbrow TCP is a probabilistic approach to predicate losses before TCP’s conventional mechanisms are triggered for congestion control. Highbrow TCP doesn’t differentiate losses but it can be used to avoid pauses in communications by redundant retransmissions. This paper is simulation based comparison of TCP NewReno and TCP NewReno with Highbrow TCP scheme. These two variants are implemented in NS-2.35 and have been tested over 3 different scenarios of MANETs. Performance is measured in terms of average throughput, instantaneous throughput and goodput with various transmission rates. General Terms Wireless Adhoc Networks, Mobile Adhoc Networks, Congestion Control, Loss Differentiation Algorithms