The difficulty in molecular simulations of complex systems comes from the fact that there exist an astronomically large number of local minima in the energy function, forcing conventional constant-temperature canonical simulations to get trapped in a few states of the energy local minima. Novel algorithms that can alleviate this multiple-minima problem are thus in urgent demand. We have been advocating the uses of the generalized-ensemble algorithms. Generalized-ensemble algorithms, which are based on artificial non-Boltzmann weight factors, alleviate this multiple-minima problem by performing random walks in potential energy space. The advantage of these algorithms lies in the fact that from only one simulation run, one can obtain not only the global-minimum state in potential energy but also various thermodynamic quantities as a function of temperature. They are thus particularly useful for free energy calculations. Multicanonical algorithm, simulated tempering, and replica-exchange method (also referred to as parallel-tempering) are well-known examples of generalized-ensemble algorithms and have been widely used in protein folding simulations. In this talk, I will present some of the latest results of our generalized-ensemble simulations in protein folding.