We point out that the expression of the cross-spectral density (CSD) of a partially coherent flat-topped beam (PCFB) given by Ge et al. [Appl. Opt. 43, 4732 (2004)] is incorrect. The results show that the M2 factor derived based on Ge's expression leads to M2 < 1 in some cases, which is physically unacceptable. A new expression of the CSD for a PCFB is given.