The influence of the polymer solubility on the separation efficiency in thermal field-flow fractionation (ThFFF) was investigated for a polymer model system of differently branched chain walking polyethylenes in five different solvents, which were selected depending on their physical parameters. The understanding of polymer thermal diffusion has been elucidated using a revisited approach based on the latest thermal diffusion prediction model by Mes, Kok, and Tijssen combined with the Hansen solubility theory. Thereby, a significant improvement in the precision of the thermal diffusion prediction and the separation efficiency has been achieved by implementation of the temperature dependency on Hansen solubility parameters. In addition, we demonstrate a method for validation of the segmental size of polymer chains with varying topology by using the revisited thermal diffusion prediction approach in inverse mode and experimental thermal diffusion data.