In fast-varying channels, an OFDM system needs to insert denser pilot symbols among transmitted symbols for tracking the variation of channel. However, using denser pilot symbols reduces the transmission throughput. In this paper, we propose a pseudo-pilot algorithm for data detection in fast-varying channels without increasing the pilot density. In our algorithm, we select some data symbols among a block of transmitted symbols, and the selected symbols are considered as pseudo-pilot symbols. Then based on a regressional model-based least-square-fitting (LSF) approach, the receiver associates each possible pattern of pseudo-pilots with a data sequence and a corresponding metric. AU possible patterns of pseudo-pilots are searched and among which we choose the associated data sequence whose corresponding metric Is minimum. The proposed algorithm needs to search all possible patterns of pseudo-pilots, and therefore the complexity may increase with the number of pseudo-pilots and constellation size. To reduce the searching number, we further propose two modified schemes. In the first modified scheme, once the metric is smaller than a threshold, we stop searching other patterns of pseudo-pilots. The second modified scheme further computes a preferable pattern of pseudo-pilots before the searching process. The simulation results show that the performance of the proposed algorithm could approach a BEP lower bound that is obtained by letting the receiver know the true values of pseudo-pilots. Comparing with the linear interpolation method, the proposed algorithm shows obvious improvement in fast-varying channels. The proposed modified schemes could effectively reduce the searching number while maintain the performance.