Correlation and prediction of miscibility in binary blends of homopolymers is investigated using an activity coefficient model, the entropic free-volume/UNIFAC (EFV/UNIFAC), and two equations of state: the Peng-Robinson (PR) and the Sanchez-Lacombe (SL) ones. Satisfactory correlation results are obtained with all models but their quality depends on whether T-dependent or T-independent interaction parameters (IPs) are used. Satisfactory prediction of the polymers molecular weight (MW) effect of blend miscibility is possible only with the EFV/TJNIFAC and SL models, while the latter predicts successfully the MW effect in the single available in the literature case for homopolymers, the blend poly(n-pentyl methacrylate)/polystyrene, where UCST and LCST as well as hourglass behavior is observed. The SL model provides also satisfactory prediction of the pressure effect on the critical solution temperature (CST) but less so on the corresponding composition. It is, thus, the most successful of the three models considered here. Finally, an empirical scheme for the prediction of UCST using the PR EoS is presented.