Location is a crucial part of wireless sensor networks technologies and applications. RSS-based (based on received signal strength) location estimations play an important role in practice. Considering the characteristic that the variance of RSS varies in different estimation points, a practical RSS-based probabilistic model is tailored and established according to the probability-based maximum likelihood in this paper. Next, taking the highly nonlinear characteristic of the object function in this probabilistic model, a location approach using the probability maximum with evolutionary algorithm (PMEA), which corresponds more to the characteristic of communication of the sensors, is proposed to find out the maximum likelihood point. The convergence is proved by the stochastic process. The results of the proposed algorithm, when implemented in a public dataset, show that this proposed probabilistic model and PMEA outperform existing solutions in terms of RSS-based location estimation accuracy.