The performance of orthogonal frequency division multiplexing systems in vehicular environments suffers from intercarrier interference (ICI) and the inherent non station- arity of the channel statistics. Receiver windowing constitutes an effective technique for enhancing the banded structure of the frequency-domain channel matrix, thus improving the effec- tiveness of a banded equalizer for ICI mitigation. However, its optimality has been verified only for stationary channels with perfectly known statistics. In non stationary channels, the second-order statistics have to be tracked and the optimal per- formance can be achieved at the expense of cubic complexity over the number of the subcarriers. To overcome this limita- tion, an adaptive windowing technique is proposed that is able to track directly an optimal receiver window in terms of average signal-to-interference noise ratio, requiring only linear complex- ity. Extensive simulation results verify both the ability of the proposed approach to track the time varying channel statistics and its increased robustness to channel estimation errors that are common in vehicular environments.