Strategy-proofness, requiring that truth-telling be a dominant strategy, is a standard concept in social choice theory. However, this concept has serious drawbacks. In particular, many strategy-proof mechanisms have multiple Nash equilibria, some of which produce the wrong outcome. A possible solution to this problem is to require double implementation in Nash equilibrium and in dominant strategies, i.e., secure implementation. We characterize securely implementable social choice functions and investigate the connections with dominant strategy implementation and robust implementation. We show that in standard quasi-linear environments with divisible private or public goods, there exist surplus-maximizing (non-dictatorial) social choice functions that can be securely implemented.