abstract: We propose a method to construct proper Lagrangian fibrations of symplectic manifolds starting from suitable affine manifolds with singularities. I shall explain in detail how this technique works in dimension two and how one can use this to construct a Lagrangian fibration of a K3 surface. In joint work in progress with D. Matessi we expect to push this method forward to obtain Lagrangian fibrations of Calabi-Yau 3-folds - e.g. the quintic - using suitable models of singular Lagrangian fibrations. I shall describe two of the singular models that can be used in this case, the so-called generic and positive models. The construction of the third model - the negative one - will be addressed in the second talk (http:/www.ma.ic.ac.uk~smscrmMatessi.html).