In photon correlation spectroscopy(PCS)particle sizing technique, the measurement accuracy is concerned with total sampling quantity N. the larger N is, the closer to true value will measured autocorrelation function be. However, large amount of sampling quantity means long measurement duration, and significant addition of memory cells is desired for software correlation in batch process mode. The reasonable sampling quantity can be determined through the estimation of relative error of autocorrelation function. Relative error of autocorrelation for correlator of M channels is approximate to±(M/N)1/2. For different particle systems and correlator channels, measurement duration ranges from several seconds to minutes. In practice, noise is another important factor concerned with the choice of sampling quantity. For the same particle system and the same number of correlator channels, measurement repeatability and deviation varies under different error level, can be improved through increasing the sampling quantity and adding correlator channels. Autocorrelation functions of simulated scattered light signal of 50nm, 100nm and 300nm particles were inversed under noise levels of 0, 0.01 and 0.1 respectively. Results show that, if there's no noise, only with small sampling quantity can ideal test result be achieved, while under noise level nonzero, to enhance measurement repeatability and minimize deviation, the desired sampling quantity need to increases with the increase of noise. Increasing sampling quantity will decrease fitting errors caused by the noises, which make the inverse algorithm produce better results.