We prove that over an arbitrary global field, for every $g>3$ there exists an abelian variety which is not isogenous to a Jacobian. We will also discuss this question over finite fields. This is joint work with Jacob Tsimerman.