Deadlocks are hard to find via traditional testing, and they manifest rarely during program execution. We introduce Lockout, a technique and a tool that increases the probability of deadlock manifestation in multithreaded programs, while preserving the program semantics and requiring no perturbation to the runtime and the testing infrastructure. Lockout produces binaries that are more prone to deadlock compared to native binaries. We evaluated Lockout on a suite of multithreaded programs, and preliminary results show that it is effective in increasing the deadlock probability.