When Segre proved his celebrated characterization of conics (``every set of $q+1$ points in $PG(2,q)$, $q$ odd, no three of which are collinear, is a conic''), he did more than proving a beautiful and interesting theorem; he in fact provided the starting point of a new direction in combinatorial geometry. In this branch of combinatorics the idea is to provide purely combinatorial characterizations of objects classically defined in an algebraic way.\\ In this talk we consider the following question: \begin{quote} Is it possible to characterize finite classical polar spaces by their intersection numbers with respect to hyperplanes and subspaces of codimension $2$? \end{quote} {\bf Remark: I do not assume knowledge of finite geometry.}