In order to study different flow patterns’ dynamic characteristics of gas-liquid two-phase flow, three time-frequency analysis methods are introduced to process the dynamic differential pressure signal of gas-liquid two-phase flow, such as the wavelet transform, Hilbert-Huang transform and adaptive optimal-kernel method. The results show that the main part of energy is transferred from frequency band 15—35 Hz to 0—8 Hz when the flow pattern changes from bubbly flow to slug flow and plug flow, and two spectrum peaks are observed at slug flow. The experimental results show that the time-frequency resolution of Hilbert-Huang transform and adaptive optimal-kernel is higher than that of wavelet transform. The extractions of ridge information based on adaptive optimal-kernel overcome the influence of fuzzy plane windowing effect, and enhance the readability of time-frequency plane information. The time-frequency analysis clearly shows the dynamic characteristics of different flow patterns, and describes the variation rules with time. It is helpful to further study the mechanism of gas-liquid two-phase flow.