Chirp pulse compression technique can improve signal-to noise ratio of ultrasonic flaw signals. But its disadvantage is the presence of sidelobes which might mask nearby small flaws and make it difficult to separate multiple flaws. In this paper, time-frequency analysis (t-f analysis) is applied to ultrasonic chirp pulse compression signals for range resolution improvement. Short Time Fourier Transform (STFT) is adopted for t-f analysis, followed by the application of Minimization and Polarity Thresholding Algorithm to suppress the sidelobe. In order to achieve the low sidelobe level without suppressing the mainlobe, the STFT is calculated by varying the window length with frequency and both pseudo-twofold and pseudo fourfold frequency cos components obtained from the STFT are utilized. The method is applied to both simulated and measured ultrasonic flaw signals of chirp. The results show a good suppression of sidelobes and a further compression of pulse width with preserving both the polarity and the amplitude of signals.