After a quick review of transplanckian-energy string collisions I will focus on a recent S-matrix approach to gravitational scattering and on its surprisingly good consistency with known results in classical gravitational collapse. A number of yet unresolved issues will also be presented.