Cubic equations of state (CEoS), such as the advanced Peng-Robinson (APR) EoS, are convenient for use in commercial simulators and have successfully fit saturation pressures and asphaltene onset points for bitumen solvent systems using simple quadratic mixing rules. However, this approach dos not accurately predict asphaltene precipitation yields. In this study, the APR EoS with several sets of asymmetric mixing rules is evaluated against saturation pressure and asphaltene yield data for n-pentane diluted bitumen. The asymmetric van der Waals, Sandoval et al., and two forms of Huron-Vidal mixing rules with an NRTL (non-random liquid theory) activity coefficient model are considered. The use of asymmetric mixing rules significantly improves the match to asphaltene yield data; however, the yields are still:underpredicted at high solvent contents, and the tuning parameters that give the best match for asphaltene yield data are not predictive or easily correlated for other solvents. The APR EoS with symmetric van der Waals mixing rules is also evaluated with compositionally dependent binary interaction parameters. The use of compositionally dependent solvent/asphaltene binary interaction parameters allows the model to fit asphaltene yield data over the entire composition range. A set of interaction parameters is recommended that fits both asphaltene yield and saturation pressure data. The merits of this methodology as a practical option for modeling heavy oil solvent behavior are discussed.