A near backscattered laser Doppler system was presented to carry out velocity and size distribution measurements for irregular particles in two-phase flows. The technique uses amplitudes of particles Doppler signals to estimate the particle size distribution in a statistical manner. Holve's numerical inversion scheme is employed to unfold the dependence of the scattered signals on both particle trajectory and orientation through the measurement volume. The performance and error level of the technique were simulated, and several parameters including the number of particle samples, the fluctuation of irregular particle response function, inversion algorithms, and types of particle size distribution were extensively investigated. The results show that the size distributions for those irregular particles even with strong fluctuations in response function can be successfully reconstructed with an acceptable error level using a Phillips–Twomey-non-negative least-squares algorithm instead of a non-negative least-squares one. The measurement system was then further experimentally verified with irregular quartz sands. Using inversion matrix obtained from the calibration experiment, the average measurement error for the mixing quartz sands with a size range of 200–560\mu \text{m} are found to be about 23.3%, which shows the reliability of the technique and the potential for it to be applied to industrial measurement.