An algorithm which uses the Kalman filter to estimate the time-varying fading coefficients in orthogonal frequency division multiplexing (OFDM) systems is addressed. The algorithm is based on a parametric channel model where the channel response is characterized as a collection of sparse propagation paths. We investigate into how to estimate the channel by the Kalman filter in the pilot-aided OFDM wireless system. To reduce the complexity of the high-dimensional Kalman filter for channel estimator, we consider a low-dimensional Kalman filter for the channel estimation. It is shown that the algorithm is effective for the estimation of the OFMD time-varying fading channels when the performance of the channel estimation is presented in terms of the mean-square error (MSE).