In IEEE 802.11 WLAN networks, the Binary Exponential Backoff (BEB) protocol has been used to avoid collisions due to channel errors, but it is less efficient for dense networks with many transmitting nodes. In this work, we propose a new scheme for adapting the contention window based on the transmission state and random backoff history. We propose four transmission states by aggregating the success or failure status of previous and current transmissions. These states are the conditions for selecting the increasing or decreasing contention window for the subsequent transmission time. The ratio between the random backoff value and the contention window partially determines the amount of change for the new contention window. We use NS-3 simulations to evaluate the performance of our scheme. The results show that the scheme can improve the total throughput and average delay and be more efficient with more nodes.