Carrier sensing has been used as an effective way to reduce collisions and exploit spatial reuse in wireless networks. Previous research has attempted to tune the carrier sensing range to maximize the network throughput. However, the impact of carrier sensing threshold on the probability of successful transmission has been ignored. In this paper, we derive an analytical model to calculate the successful transmission probability. We then calculate the throughput of routing protocols using different link metrics. To the best of our knowledge, this perhaps is the first attempt to derive the throughput of routing protocols like expected transmission count (ETX)  and expected transmission time (ETT)  in wireless networks. We also investigate the impact of some other important factors, such as node density, average contention window size and packet length. Our results show that optimal routing protocols that are using ETT as the path metric can achieve around 30% more throughput than those using ETX and End-to-end delay. Compared to the minimum hop count protocols such as DSR and AODV, the optimal routing protocol can improve the throughput by up to 100%.