The Constrained-Routing and Spectrum Assignment (C-RSA) problem arises in the dimensioning and management of a next-generation of optical transport networks, called Spectrally Flexible Optical Networks (SFONs). The C-RSA can be stated as follows. Given an SFONs as a graph G, and an optical spectrum S of available contiguous frequency slots, and a multiset of traffic demands K, it aims at determining for each demand k ∈ K a path in G and an interval of contiguous slots in S while satisfying technological constraints, and optimizing some linear objective function(s). To the best of our knowledge, a cuttingplane-based approach has not been yet considered for the problem. For that, the main aim of our work is to introduce an integer linear programming formulation and provide several classes of valid inequalities for the associated polyhedron. We further discuss their separation problems. Using the polyhedral results and the separation procedures, we devise a Branchand-Cut algorithm to solve the problem. We also present some computational results and show the effectiveness of our approach using real and some realistic network topologies.