Let X be a surface, possibly with boundary. Suppose it has infinite genus or infinitely many punctures, or a closed subset which is a disk with a Cantor set removed from its interior. For example, X could be any surface of infinite type with only finitely many boundary components. We prove that the mapping class group of X does not satisfy the Tits Alternative. That is, Map(X) contains a finitely generated subgroup that is not virtually solvable and contains no nonabelian free group.