An orthogonal frequency-division multiplexing (OFDM) system with non-uniformly spaced pilots, such as worldwide interoperability for microwave access, voids virtually all discrete Fourier transform (DFT)-based interpolation methods for channel estimation, because they all require pilots to be uniformly spaced. Therefore, we propose two iterative algorithms to solve this problem. In the proposed algorithms, the inverse discrete Fourier transform and DFT are iteratively used to smooth the channel frequency response estimated by linear interpolation (LI) or other means. To be complete, we also propose a method to truncate the channel impulse response for the proposed algorithms. The simulation results show that the proposed approach has much lower mean square error in a long delay channel than the LI method has, and it also outperforms a decision-directed approach at high speed and/or lower signal-to-noise ratio environments. Overall, the proposed approach is a promising approach for channels with long delay spread in an OFDM system with non-uniformly spaced pilots.