A recent paper in this journal by Deng, Li and Li has investigated methods to estimate rates and effects of polygenic mutations using data from mutation accumulation experiments. Here, I evaluate a number of critical points in this paper concerning a maximum likelihood (ML) procedure to analyse mutation accumulation data. I show that Deng, Li and Li's criticisms are based on misunderstandings, or numerical problems they encountered that could have been readily overcome. In Monte Carlo simulations, I show that ML can give a considerable increase in precision over the method of moments that is traditionally used to analyse mutation accumulation data. Furthermore, ML allows the comparison of the fit of different models for the distribution of mutation effects.