Compressive sensing (CS) is implemented to reconstruct the particle size distributions. For unimodal distributions, sampling rate has little relation to reconstruction quality when it is beyond a certain level. The proposed method is effective and robust for particle sizing from noise-contaminated data.