Abstract: A comparison of three PIV algorithms: FFT-based cross-correlation (FFT-CC) method, direct cross-correlation (DCC) method, and minimum quadratic difference (MQD) method was performed by applying these methods to low number density PIV images obtained by the Japan Visualization Society.Raw results of the MQD method were better than those of the FFT-CC, but validated results of the FFT-CC method were better than those of the MQD method. Results of the DCC method were the worst.There was discrepancy between results of the same algorithm with different sizes of the interrogation window.It mainly resulted from various amounts of matched particles of sub images.This study could provide good reference to the selection of the algorithm and its parameters.