Time-frequency (TF) analysis (TFA) can reveal geological information hidden in seismic reflection data, but most methods focus on trace-by-trace analysis of seismic data with a lack of lateral constraints. In this letter, a novel multichannel TF transform (MCTFT) is proposed and applied for describing tight sandstone channel reservoirs. First, a formation dip parameter is introduced to capture the lateral variation of seismic data. Then, the MCTFT is constructed with the windowing idea of short-time Fourier transform (STFT), where the dip parameter is determined by solving a plane wave equation. Moreover, the coherence is taken as a trade-off between single- and multichannel processing to avoid the latter over-smoothing the edges of geological features. Compared to the single-channel TFA method, the MCTFT can provide a more laterally continuous spectral decomposition that clearly describes tight sandstone channel reservoirs. A synthetic wedge model and two field seismic data are employed to verify the effectiveness of the proposed method.