<p>Differential evolution (DE) algorithm is a famous representative of swam intelligence methods. It is good at continuous optimization problems but poor at discrete optimization problems. To overcome the shortage and take advantage of its continuous evolution, a continuous differential evolution (CDE) algorithm is proposed to solve uncapacitated facility location problems (UFLP). In CDE, firstly, based on opposition-based learning technique, an initial population is randomly produced in the range of [0,1]. Secondly, a mechanism of perturbation of fixed number is integrated into its crossover operation. Thirdly, a random replacement strategy is proposed to repair infeasible individuals to guarantee that at least one component is in the range of [0.5,1] for each individual. Fourthly, a probability discretization mechanism is presented to map a continuous individual to a binary vector solution, which can be used to calculate objective function of UFLP. Thus, CDE can use original operators of DE to evolve in continuous space. In addition, it can also solve binary UFLPs. Lastly, 35 famous benchmark instances of UFLP are used to test the performance of CDE. Moreover, it is also compared with 16 state-of-the-art algorithms on these benchmark instances. Experimental results show the performance, simplicity, superiority and robustness of CDE.</p>