I will review the PBL representations in GCMs, especially the mixed-layer approach used in UCLA GCM. The review will start with a statement from Arakawa that 'Around 1970, I had the feeling that parameterization of PBL processes was one of the weakest aspects of the GCMs that existed at that time' and end with another statement that `I still feel that, in view of their crucial roles in the climate system, the formulation of PBL processes remains among the weakest aspects of the existing GCMs. This is especially true for the stratocumulus-topped PBL.'