Small changes of parameters of inversion algorithm for dynamic light scattering will result in large deviation of its solution. In order to improve the measurement accuracy, it is need to establish a set of criteria to evaluate the solution of inversion algorithm. The paper analyses the inversion algorithms in three aspects: fitting precision of correlation function, stability of particle size distribution, and repeatability of solution, then proposes three evaluation criteria. Criteria I: root mean square　(RMS) error of fitting correlation function should be smaller than 0.001, quality factor(Q) of error should be Q>0.7; criteria II: sample standard deviation (RN) of the norm of particle size distribution should be RN<5%; criteria III: relative standard deviation (RSD) of the solution should be smaller than 2%. Experimental data show that in case of the inversion algorithm meets all three evaluation criteria, it can be obtained the most close to the actual measurement results with good stability and high repeat precision.