The predictive local composition model is applied to multicomponent hydrocarbon systems with long-chain n-alkanes as solutes. The results show that it can successfully be extended to high-order systems and accurately predict the solid appearance temperature, also known as cloud point, in solutions of known composition using only binary information. The model can describe the experimentally well-known capacity of heavy alkanes to act as cloud-point depressants and explains it in terms of the nonideality of the solid solutions formed. There are good indications that no significant error is introduced by considering the multiple orthorhombic phases as a single phase. It is shown that for low concentrations in heavy fraction the interaction parameter alpha(1s) can be neglected and thus the model becomes purely predictive.