Mixed-integer nonlinear optimization problems involving polynomials can be reformulated by creating an auxiliary variable for every monomial and convexifying the product constraint of each such variable. This gives rise to multilinear polytopes. In the talk I will describe this relationship, summarize what is known about these polytopes, and then focus on known inequalities classes. The emphasis will be on separation algorithms in order to use these inequalities as cutting planes. Finally, I will present some insights from a prototype implementation. The presented new results are joint work with Alberto Del Pia.